What to consider about events: A survey on the ontology of occurrents

2.1. Employed terminology

In this paper, we will employ the term occurrent to refer to the most general type of thing that happens in time (usually referred to as “perdurant”, “occurrence”, or simply “event”). Analogously, we will employ the term continuant to refer to the most basic type of thing that exists in time and endure change (which is called “endurant” or even “object” in some ontologies). We will use prolonged occurrent to refer to occurrents that happen over a non-instantaneous interval (e.g., the writing of a book) and instantaneous occurrent to indicate occurrents that are restricted to a single instant of time or that are temporally non-divisible at the considered granularity level (e.g., the ignition of a fire). Temporal region and time instant/point will be preferred to refer to the position or distribution of things in time.

The term object will be used to refer to existentially independent continuants (e.g., a person, a rock), while quality will be used to represent the intrinsic features presented by independent continuants at some point in time – both those representing valued attributes (i.e., color, size) or representing conditions, dispositions and the like (i.e., a fever, magnetic nature). Spatial region and spatial point will be employed to locate things on the 3-dimensional space, with spatiotemporal region and spatiotemporal point being their analogous in 4-dimensional space.

The term situation will refer to some configuration of a considered portion of the world at some spatial region and time point or interval (i.e., the set of objects that are present in such spatiotemporal region, along with the properties they present and relationships established among these objects, e.g., a prepared patient and a surgical team on an operating room right before a surgery). The term role will be used to refer to the common sense notion of role, as those contingent types that an instance may assume in some context, such as relational roles (e.g., student), occurrent-bounded roles (i.e., roles that an entity play with respect to an occurrent, e.g., agent), or roles of some other nature. The term relation will be employed to refer to something that holds between entities, connecting them in some sense (for which sometimes it is used the terms “relationship” or “object property”), and the term relationship will be regarded in a broader sense (e.g., “the relationship between these two notions”). We will resort to the term concept to broadly refer to the notion of a set of properties which form an ideal model that can be concretely exemplified in the world (for which sometimes it is used terms like “universal” or “class”). Complementarily, we will employ the term instance to refer to concrete exemplifications of concepts (for which sometimes it is used terms like “particular” or “individual”). Additionally, we will use the terms type and subtype to emphasize the idea of taxonomic division (e.g. “types of occurrents” would mean the various specializations of the concept of “occurrent”).

2.2. Considered Ontologies

There is a considerable number of ontologies that deal, at least in part, with the issue of occurrents. Then, bearing in mind that the focus of the paper is not assessing different ontologies, but rather bringing together remarkable ideas about the representation of occurrents (which may help to solve practical modeling questions, or face questions on the very nature of occurrents), we made our best efforts to have a sample of ontologies that is representative, but still small enough to allow treating the noteworthy aspects of occurrents in detail. Thus, since we are interested in the most general aspects involving occurrents, we avoided considering domain ontologies of any kind (e.g., CIDOC1

The Simple Event Model (SEM; van Hage et al., 2011, 2009) is a model/ontology for representation of occurrents on the web, intended to achieve interoperability with data sets from various domains. It is designed with a minimum semantic commitment, assuming nothing about the data sets that will be interoperated. The model only provides a single concept of occurrent (i.e., Event), but allows the representation of further types. It also provides means to represent place and time of occurrence and the involved participants as well as their roles. Also, being devised to go “beyond purely ‘factual’ occurrent models” (van Hage et al., 2011), SEM brings the notion of views according to which occurrents are considered.

The Event and Situation Ontology (ESO; Segers et al., 2015; Segers & Rospocher, 2015) is designed for extracting implicit information from text. ESO does not provide a complete definition of events and situations, but allows the formalization of pre, during and post-conditions of an occurrent, as well as the roles of entities affected by the occurrent. It proposes three types of occurrent (i.e., Event, Dynamic Event, and Static Event) and reuses other concepts from SUMO, WordNet, and FrameNet (that will not be considered in this survey).

The Versatile Event Logic (VEL; Bennett & Galton, 2001) is a highly expressive language for representing temporal relationships and occurrents, based on an ontology of occurrents (which we will refer to as “VEL”) accounting for their various linguistic manifestations. It allows the representation of events and temporal relationships, but it is not intended to directly contribute to reason about actions and changes, describing a few occurrent types (e.g., Punctual Event vs. Durative Event, Unitary/Solid Event)

The Event-Model-F (Scherp, Franz, Saathoff & Staab, 2012) is a pattern-oriented ontology for capturing and representing occurrents in the real world, in order to allow easy interchange of information about occurrents between different occurrent-based systems. Event-Model-F is based in DOLCE+DnS Ultralite (DUL)5

The ontology presented in Kaneiwa, Iwazume & Fukuda (2007) – which we will refer to as “KAN” – is an upper ontology for occurrents proposed as an infrastructure for event knowledge bases and intended to provide help in tasks such as annotation of event data and detection of relations between events. KAN describe occurrents both according to their component structure (e.g., the number of participants involved and the roles they play) as well as according to their semantic function (i.e., the effect of the occurrent over its participants). It also defines types of relations between events, both in instance-level and type-level. KAN proposes 20 types of occurrents, 14 of them concerning the component structure of the occurrent (e.g., natural event vs. artificial event, dynamic state vs. static state) and the other six concerning semantic functions (e.g., state change, temporal existence change, object identification change).

The Descriptive Ontology for Linguistic and Cognitive Engineering (DOLCE; Gangemi et al., 2002; Lehman et al., 2004; Masolo et al., 2003; Laboratory for Applied Ontology, 2005) is an upper-level ontology of particulars. As the name suggests, DOLCE has a clear cognitive bias, aiming to capture the ontological categories underlying natural language and human common sense. Thus, its categories reflect the surface structure of language and cognition, being regarded as cognitive artifacts, not necessarily reflecting the intrinsic nature of the world. Concerning occurrents, DOLCE introduces nine concepts (e.g., event, state, process, activity, phenomena).

YAMATO (“Yet Another More Advanced Top-Level Ontology”) (Mizoguchi, 2010; Galton & Mizoguchi, 2009; Mizoguchi, 2012) is an upper ontology intended to give a better treatment to three main ontological “dimensions”: quality and quantity description, ontology of representation, and processes and event modeling and distinctions. It assumes a Newtonian, mesoscopic world and adopts a view in which continuants (particularly, objects) and occurrents are mutually interdependent, being neither of these categories ontologically prior to the other. Concerning occurrents, YAMATO brings about 200 concepts.

The General Foundational Ontology (GFO;6

The Basic Formal Ontology (BFO; Grenon & Smith, 2004; Smith & Grenon, 2004; Masolo et al., 2003) is a foundational ontology designed to reconciling three-dimensionalist and four-dimensionalist views. Thus, it’s composed of two “modules” representing two different perspectives of reality – SNAP and SPAN. SNAP deals with three-dimensional objects, their qualities and the spatial regions they occupy, being used to characterize static views of the world. SPAN is an ontology of occurrents as the things that persist in time through their temporal parts, such as the processes and the spatiotemporal regions in which they occur. Regarding occurrents (considered here as happenings, thus excluding spatiotemporal regions), BFO provides about six concepts.

The Suggested Upper Merged Ontology (SUMO; Niles & Pease, 2001; Pease, 2018) is an extensive formal public ontology that follows pragmatic principles, excluding distinctions of strictly philosophical interest. It is composed by a top-level ontology, a middle-level ontology, and various domain ontologies. Regarding occurrents, SUMO provides about 1200 concepts – the majority being provided by domain ontologies.

The Unified Foundational Ontology (UFO; Guizzardi, 2005; Guizzardi et al., 2007; Guizzardi & Wagner, 2010; Guizzardi et al., 2013; Guizzardi et al., 2016) is a philosophically and cognitively well-founded reference ontology, encompassing both continuants and occurrents. It is subdivided in a series of partial and interconnected ontologies devised to deal with different views of reality. The main ones are UFO-A (an ontology for continuants) and UFO-B (an ontology for occurrents). UFO-B covers aspects such as mereology of occurrents, temporal ordering, and participation of continuants, and involves over a dozen concepts for occurrents.

3.1. Definitions of occurrent according to different ontologies

In view of discussing the definitions of occurrent proposed by each ontology, we may consider the very loose and hopefully uncontroversial (at least with respect to the ontologies reviewed in this paper) definition of occurrent earlier presented (i.e., anything that happens in time) and, starting from that, go on adding details gathered from the definitions proposed by each ontology.

Indeed, this definition is very close to that proposed by SEM, which states that an occurrent (represented by the concept of event) encompasses everything that happens – including even fictional events (e.g., King’s Arthur quest) – disregarding whether or not they happen in a specific place or time, or whether there are specific actors involved (although the simple mention to those characteristics may reveal some sort of commitment to such properties of occurrents). SUMO also acknowledges the concept of process (its topmost concept related to the idea of occurrent) as anything that occurs in time but is not an object (its topmost concept related to the idea of continuant, though only referring to independent continuants), that has temporal parts or stages and that may also have objects as participants.

DOLCE goes on stating that occurrents (which, at the instance level, are referred to as occurrences or perdurants, with the term occurrent denoting the corresponding universal) happen in time by accumulating different temporal parts. Due to that, at any time they are present, they are only partially present, since some of their proper parts (e.g., past or future phases) will not be present. According to DOLCE, occurrents cannot change in time. Thus, if an occurrent presents some property at a time at time t 1 and present an incompatible property at time t 2 , it is not the case that the occurrent has changed – it is just two different temporal parts of it exhibiting different properties. For example, if a talk is boring during its first half and becomes interesting later on, the talk suffered no change, it was always the same: a talk whose first half is boring and last half is interesting.

In DOLCE, the concept of occurrent comprises those of state (i.e., occurrent that is both cumulative7

KAN seems to emphasize the change nature of the notion of occurrent (represented by the concept of event), defining it as something dynamic, that affects an object or an environment, changing their qualities in the next time. In the case of affecting environments, the occurrent is represented without a clear definition of the object participants involved. In spite of this dynamic nature, they also propose the concept of static state as a type of occurrent, for representing states such as “hot”, “cold”, and “fine”. This notion of static state seems close to the notion of qualities of continuants, since they are even regarded as able to undergo change (e.g., the dynamic state – another type of occurrent in the ontology – of “becoming higher” indicates a change on the static state of “low temperature”, representing that the temperature of this state will get higher). Nevertheless, static states could be regarded as occurrents if we consider it as the process of maintaining a state (e.g., “being hot” would be the occurrent of actively keeping the high temperature of something) – which is not clearly stated in the ontology, though. KAN characterizes an occurrent both by its component structure and its semantic function. The component structure of an occurrent comprises the relationship among its components, which includes participants, time and location. Semantic function represents the types of change caused by the occurrent on the truth values of properties and states over dynamic times and locations. Besides the distinction between dynamic state (which implies activity or dynamic change of an object or environment) and static state (which implies a static property), another important distinction in this ontology is that between artificial event and natural event, (regarding the participation of some agent on the occurrent).

According to the ESO, an occurrent (represented by the concept of event) is also an entity that describes some change or state in the world, being associated with some participants, a time interval and, typically, three situations: a pre-situation (i.e., that holds before the event), a during-situation (i.e., that holds while the event is happening), and a post-situation (i.e., that holds after the event). Also, ESO accepts that an event exists regardless of whether it happened or not – thus, including hypothetical events, for example. Analogously, VEL follows the same idea, explicitly defining an occurrent (or event, as it is called on the ontology) as a transition between world states – with “world state” being regarded as a set of continuants with their particular properties at a time instant, in resemblance to the notion of situation used in this paper. Again, occurrents in VEL also have continuant participants and are correlated with the time intervals over which they occur. As world states may represent similar configurations of the world, it seems possible to accommodate in this ontology a notion similar to that of “maintaining a state”, although it is not explicitly presented as a type of occurrent.

UFO also commits to the idea of occurrents (initially represented by the concept of event) as things that happen in time, accumulating temporal parts, that represent changes of the state of affairs from one situation to another. Furthering this definition, it makes explicit the nature of such change, defining events as manifestations of individual qualities, specially dispositions (i.e., properties that are only manifested in particular situations, and that can fail to be manifested, e.g., propensities, capabilities, liabilities). It makes events existentially dependent on such qualities and therefore also (indirectly) existentially dependent on the continuants that bear these qualities (Guizzardi, Guarino & Almeida, 2016).

In a sort of follow up work on the conceptual basis of UFO (Guarino & Guizzardi, 2016) (which, for the best of our knowledge, is not yet included on the ontology), the authors propose that the prior concept event would be solely a subtype of occurrent, introducing the concept of scene (i.e., whatever happens in a suitably restricted spatiotemporal region – the maximal occurrent on such region) as the complementary type. Following from that, they enrich the definition of event as whatever happens to a selected set of individual qualities of objects in a particular spatiotemporal region. This set of qualities would be the focus of the event, with different foci yielding different events, so that it would work both as a principle of individuation and an identity criteria. Therefore, scenes could be roughly defined as “occurrents without focus” and events as “occurrents with focus”, with events emerging from scenes through a process of focusing on a selected set of individual qualities of one or more objects. Finally, according to UFO, occurrents cannot change in time. Such as in DOLCE, if an occurrent presents some characteristic at a time and presents a different characteristic at another instant, it is just two of its temporal parts presenting different properties.

For BFO, an occurrent (represented by the concept of processual or processual entity) is a four-dimensional entity that occurs in time and unfolds itself through a time interval in its successive temporal parts. As occurrents only exist in their temporal parts (or phases), they are bound in time, so that each portion of the time during which an occurrent occurs can be associated with a corresponding temporal portion of the occurrent. Moreover, once again, as in DOLCE and UFO, occurrents cannot bear changes. Occurrents are existentially dependent on continuants as their bearers and, again similarly to UFO, represent manifestations of dispositions/qualities. Occurrents are not located in space, but in spacetime (whose instantaneous slices correspond to space regions). Among others, BFO makes distinctions between prolonged occurrents (i.e., processes) and instantaneous occurrents (i.e., process boundary or instantaneous temporal boundary, used to represent both fiat and bona fide boundaries of prolonged occurrents – such as their beginnings and endings – as well as instantaneous transitions within prolonged occurrents).

YAMATO also adheres to the idea that occurrents (represented by the occurrent concept) are things that extend in time, having temporal parts, and that can represent changes. Nevertheless, it also includes as types of occurrent the concepts of state, as the state possessed by an entity, and situation, as a state of affairs or interaction pattern between continuants. As in KAN, such concepts could also be regarded as occurrents that represent the maintenance of the corresponding states or situations (in the sense adopted in this paper to the notions of state and situation). Moreover, being a change, an occurrent must be a change of something – so that, every occurrent must have some material support, which generally refers to independent continuants, in a broad sense, so as to include, for example, energy and fields of force. This implies the idea of having participants and could allow deeming occurrents as dependent on continuants. However, unlike previous views, which seemingly assume continuants as the main components of the ontology and occurrents as dependent on them, YAMATO has no ontological priority between continuants and occurrents, being both mutually dependent. The ontology present objects as interfaces between the occurrents that represent their external processes (i.e., the behavior presented by the object) and the occurrents representing their internal processes (i.e., those that sustain the existence of the object, collaborating to the occurrence of its external processes).

In YAMATO, there is an important distinction between two types of occurrent: processes and events. A process is regarded as ongoing stuff, having no whole and no temporal parts (e.g., the process of “running”). On the other hand, an event is a unitary whole with respect to time, representing a distinct individual episode with a definite beginning and end (e.g., a run). In other words, whereas processes are non-discrete, dissective (i.e., any part of a process is itself a process of the same kind as the whole), having no definite extension, events are discrete, non-dissective and with a definite extension. In addition to that, the relationship between events and processes is analogous to that between objects and matter, so that events are constituted by processes as their material, being “chunks” of processes delineated by boundary events (e.g., start, end) – except for those very events that act as boundaries. Following form that, analogously to the case of objects, that can vary in complexity (e.g., a piece of wood vs. a table), events can vary from simple process chunks (e.g., a hike) to complex arrangements of qualitatively distinct process-chunks ordered in a certain way to qualify as a given type of event (e.g., an exploration journey).

Finally, in YAMATO there are certain types of occurrent which may change in time – namely, processes. In this ontology, a process is not a whole and have no parts and seems to be treated as a pattern of occurrence that can be framed in a “time window”. Such time window represents a suitable time resolution corresponding to a time length in which one can observe the whole pattern unfolding (e.g., for a “walking” process, the time window would be such that it would be possible to observe full steps, not just the isolated movements of a single leg). Thus, as such time window progress in time, we observe different stages of the same process (in a way analogous to that we would observe different time slices of a continuant), which may present differences in the occurrence pattern in relation to earlier stages (e.g., the speed of walking may increase in the next stage).

GFO also commits to the concept of occurrent as something that happens in time, having temporal parts, in a sense similar to that of “occurrent” as defined in this article, except for including concepts that resemble situations (i.e., configuroids and situoids). Nevertheless, this ontology establishes that processes (a concept analogous to that of prolonged occurrents) are independent individuals of reality, with presentials (roughly speaking, the instantaneous temporal slices of continuants, that are wholly present at a time instant) existentially depending on processes. Based upon that, processes are defined as coherent temporal complexes, in the sense that their boundaries (i.e., the set of time slices of the continuants participating in a process) are ontically9

GFO presents the concept of material process as a process whose boundaries are “material structures” (i.e., temporal slices of objects/independent continuants), in opposition to the concept of quality process (i.e., whose boundaries are simply aggregate of qualities). It also brings the concepts of change (i.e., transition between two processual boundaries that present contradictory properties), discrete/instantaneous change (i.e., a change in which the contradictory processual boundaries occur one immediately after the other), and continuous change (i.e., in which the contradictory processual boundaries are apart in time). Following from that, it presents the concepts of continuous process (i.e., process with no discrete changes within it) and discrete process (i.e., having discrete changes). As a last note, GFO also presents the concepts of event (i.e., a processual boundary that presents some property that is not verified either on the immediately previous or the immediately following processual boundary) and action (i.e., corresponding to the idea of an occurrent that is caused by an agent).

Summarizing this section, we can divide the aspects included in the definitions provided by each ontology in the three groups proposed in the introduction: (i) Pervasive Aspects, (ii) Complementary Aspects, and (iii) Conflicting Aspects.

Concerning (i), all the analyzed ontologies commit to a notion of occurrent that is disjoint from that of continuant,11

With respect to the complementary aspects (ii), most of them are related to the inclusion of specific subtypes of occurrents. For example, there are occurrents having agents or not, having or not material structure, and having or not focus (which is noticeable both in the division between scene and event in UFO, as well as in the division between events that act upon a diffuse environment or upon specific objects, as proposed in KAN). We also have instantaneous or prolonged occurrents, counterfactual occurrents (such as the fictional ones in SEM or hypothetical in ESO) and those that represent an obtained result (as the event in GFO). The other complementary aspects represent modeling ideas that might help to enrich the ontological analysis of occurrents. Among them, we can mention the views of occurrents as transitions between situations, the coherence of processes in GFO (i.e., the causal and ontical connections between temporal neighbor slices of a participant), and the constitution of structured occurrents by “amorphous” ones (i.e., the relation between events and processes in YAMATO). We may also include the nature of occurrents as manifestation of individualized qualities (which is explicitly stated in UFO and BFO, but that may have some traces of it on the definition of purely quality processes in GFO), even though, in some sense, it may contradict the occurrent priority in GFO, since occurrents as manifestations of qualities are dependent on such qualities and then indirectly dependent on their continuant bearers.

To conclude the section, there are the conflicting aspects (iii), which are not compatible among themselves and could not (or at least hardly) coexist on the same ontology. Thus, they roughly correspond to different views on reality as a set of excluding options from which the modeler should choose according to her/his own view and the needs of the domain in hand. Among them, we can highlight the possibility of change for occurrents12

4.1. Participation according to different ontologies

VEL presents a rather intricate view on participation. First, a continuant instance is represented by a series of time slices of such instance, one slice for each time point of its existence. Each of such slices is referred to as the “entity” that realizes the individual in the world state that obtains at the corresponding time point. Along with that, occurrents are described using episodes (roughly speaking, the set of world states obtaining in each time point during the time interval of the occurrent). Thus, an occurrent is interpreted as the transition through such world states – each state composed by possibly many entities, that realize the different continuant instances present in the state. When a continuant participates in an occurrent, it goes through a series of changes that satisfy the conditions that constitute the meaning of the occurrent. So, the participant is modeled as a sequence of entities representing the sequence of property changes that the participant goes through during the occurrent happening. Therefore, a continuant instance participates in an event occurrence iff each state on the episode that describes the occurrent contains the corresponding entity realizing the continuant in that state. In other words, VEL commits to a restrict notion of participation in which the continuant participates during the whole occurrent.

On the other hand, DOLCE introduces time-indexed participation, so that it is possible to indicate the time interval in which a continuant participated in an occurrent. This feature yields different types of participation, such as constant participation (i.e., the continuant participates during the whole temporal extension of the occurrent) and temporary participation (i.e., the continuant participates during just a part of the temporal interval of the occurrent). Additionally, there may be temporary total participation (i.e., the continuant temporarily participates in the occurrent, in every non-temporal part of the occurrent – e.g., in every spatial part of the occurrent during the interval of participation) and total participation (i.e., the continuant participates throughout the whole temporal extension of the occurrent, in all its non-temporal parts). Related to this idea, DOLCE also includes the concepts of maximal participant (i.e., continuant that totally participates in an occurrent, during its whole existence – so that the interval of the existence of the continuant coincides with the temporal interval of the occurrent) and life (i.e., the mereological sum of all occurrents in which the continuant totally participated).

YAMATO commits to a notion of occurrent which requires the participation of some material host (understood as an independent continuant, or an object), representing the notion of participation by means of the enactment relation. In this ontology, objects (in this case, understood as delimited portions of matter) enact instances of both event and process, while matter only enacts instances of process. This is due to the fact that YAMATO events are anchored in space and time. Thus, any matter involved in an event would take the form of a definite and delimited quantity of matter, which would therefore be understood as an object or a collection of objects (e.g., the process of raining is enacted by water, but the event of a rainstorm is enacted by a particular aggregate of raindrops falling from the clouds).

YAMATO makes a distinction between the notions of external and internal process, defining that objects enact their external processes and are “supported by” their internal processes (e.g., a windmill converts energy into useful work as its external process and has the rotation of its vanes as an internal process). Based on this distinction, the ontology poses additional constraints over the notion of participation. For example, if an object is enacting a process, no proper part of the object is enacting that process (e.g., a human walks, but her/his legs do not – although the legs certainly enact some internal process that contributes to the overall walking). Following from that, processes are enacted by a single individual at any one time. Therefore, if a process seems to be enacted by several different individuals, it is in fact enacted by a composite individual which has those individuals as parts – each of them enacting its own partial process that contributes to the overall process. For example, if two people are pushing a heavy table, the manipulation of the table by each of these people are internal processes of the composite formed by the two people, whose external process is pushing the whole table. Moreover, this constraint is even proposed as unity and identity criterion for objects (i.e., object is a unity which is what enacts its external process, being an interface between its internal and external processes).

Still based on the division between internal and external processes, YAMATO also acknowledges, in a somewhat peculiar way, the possibility that an object plays a role in a process. Any object involves external processes that the object enacts and some internal processes that contribute to the enactment of such external process (e.g., a vehicle has “moving” as its external process and “providing motion power” as one of its internal processes). Each internal process of the object is specified by a role, which forms part of the definition of the object (e.g., “providing motion power” is associated with the role “power source”). Each internal role is filled by a role-holder object which enacts the process specified by the role. The role-holder may be internal or external to the object itself – if internal, it is a component; if external, an auxiliary object (e.g., for a car, the role-holder of “power source” is internal – the engine – whereas for a bicycle it is external – the person riding the bike). With that, a role-holder participates in the external process of the main object playing its internal role. In other words, it plays an internal role in the context of the external process. Finally, the processes which determine the nature of an object (i.e., its external processes) are causally dependent on the internal processes of the object. Due to this causal dependence, the object could not exist as it is without its internal process (i.e., internal processes sustains the object by giving rise to the external processes that characterize it as such). Therefore, according to YAMATO, besides enacting external processes and events, objects also relate to occurrents by being sustained by their internal processes and events.

ESO also offers support for both participation and definition of roles played in the context of an occurrent. It provides a hasRole relation for indicating the participants of an occurrent (e.g., “purchase X” hasRole “customer Y”) that can be specialized to represent the specific role a participant played in the context of the occurrent (e.g., “purchase X” hasBuyer “customer Y”). The support of SEM for participation and occurrent-bounded roles follows a similar approach. It provides a hasActor relation to indicate the participants of an occurrent (e.g., “soccer match X” hasActor “soccer player Y”). For the representation of roles, it provides the concepts of Role and RoleType, as well as the roleType relation. Then, to say that the “soccer player Y” is a goal scorer, we create an instance “goal scorer” of the type RoleType, an instance “r” of the type Role, and states that “r” rdf:value “soccer player Y” and that “r” roleType “goal scorer”.

Moreover, SEM explicitly accepts that instances of its four core classes (i.e., actor, event, place, and time) may play a role with respect to occurrents. This opens the possibility that not only continuants but also places, times and occurrents can play occurrent-bounded roles. Finally, neither ESO nor SEM provide further specifications about the semantics of the way an actor participates (e.g., in an active or passive way) or the possible types of roles, which must come from some domain ontology.

In Event-Model-F, we represent participation using the participation pattern. With this pattern, participation is expressed as a situation (F:EventParticipationSituation) that includes an occurrent (classified by the metatype F:DescribedEvent) and the continuants participating in it (classified by the metatype F:Participant). So, applying this pattern, it is possible to select an occurrent concept and a continuant concept from a domain ontology and state that instances of the former have instances of the later as their participants. Then one can instantiate such a situation so as to refer to the specific instances involved in a specific participation. The pattern defines just the participation structure, leaving to domain ontologies the definition of the roles that participants may play. With that, the domain concepts to be classified by the metatype F:Participant can be roles defined in some domain ontology (e.g., a disaster ontology may define “Disaster Subject” as the role of someone affected by an event). The pattern is complemented by parameters to describe the general spatial region where the participants are located (i.e., F:LocationParameter) as well as the general temporal region in which the occurrent happens (i.e., F:TimeParameter).

In KAN, occurrents are represented as n-ary predicates corresponding to the component structure of the occurrent. Such predicates have as their possible arguments the participants, the time and the spatial location of the occurrent. Regarding the inclusion of participants on the description of the occurrent, KAN considers two types of occurrent. On the one hand, we have object events, whose definitions include the participation of continuants (e.g., a volcanic eruption). On the other hand, KAN also presents environment events, which are defined without determining the participants involved, as occurrents that happen to an environment as a whole, with no agent or main component object (e.g., an earthquake). Among object events, we can identify four types of participants – which could be regarded as the possible roles played by continuants in relation to an occurrent. The first type is object, representing the continuant which suffers the effects of the occurrent (e.g., the patient in a medical intervention). The second type is the agent of an occurrent, which represents the participant that carries out an action over an object participant (e.g., the medic in a medical intervention). The third type could be called intransitive-agent, corresponding to the agent in an occurrent that is an intransitive action (i.e., in which the agent both carries out and suffers its own action, e.g., a breath, a jump). The last type is that of agent-object (present in group actions), in which the participant acts upon another participant and, simultaneously, suffers an action carried out by another participant (e.g., the role of a person in the event of a hug or of a discussion).

There are several forms of participation in GFO. The most evident ones are those that relate presentials (i.e., time slices of a continuant) and persistants (i.e., universals whose instances are presentials and that, roughly speaking, represent continuants in their whole temporal extension) to processes (i.e., prolonged occurrents). When a persistant participates in a process, this participation originates a layer of the process that captures the role of the participant in the process (i.e., the manner in which the participant behaves in the context of the process). Such a layer is called a processual role and is regarded as a process itself. There is a role-of relation (i.e., a special case of the part-of relation) between a processual role and its contextual process (i.e., the one from which the processual role originates). A plays relation is established between the persistant and the processual role originated by the participation of the persistant in the contextual process (i.e., the overall process of which the processual role is a part). Most processual roles are relational (i.e., they exist in relation to a distinct contextual process, that is composed of multiple processual roles, e.g., in a basketball match, the participation of each of the ten players originates a processual role). However, processes that comprise only a single participant (i.e., a person blinking her/his eyes) are simultaneously a processual role and its own contextual process.

In GFO, presentials may refer to time slices of material structures, amounts of substrate, and qualities. Thus, since GFO states participation as a relation between any presential and a process, it commits to the participation of both independent and dependent continuants – what is also observable on the definition quality process (i.e., process with any of its boundaries presenting an aggregate of qualities). As a last remark, the paper that presents GFO also mentions the participation of processes in other processes by means of the particproc relation. According to it, a process x would participate in a process y if x is a layer-part of y – which seems to make this relation close to that of parthood (that is discussed in Section 5).

In UFO, participation is also a formal relation linking continuants and occurrents, and when a continuant participates in an occurrent, it gives rise to a participation (i.e., an occurrent that exclusively depends on a single continuant). The nature of such participation induces the participant to play a processual role in relation to the overall occurrent, which is in UFO as the way in which the continuant participated in the occurrent.

In SUMO, the definition of process (i.e., occurrent) acknowledges that it may have objects (i.e., independent continuants) as participants. Such a relationship is represented by the caseRole relation, which relates a process to its participants. SUMO also provides a way of representing the roles of the participants of a process, using specializations of the caseRole relation. Those specializations include the roles of agent (i.e., the participant, either animate or inanimate, that realizes the action of the process, with or without voluntary intention) and patient (i.e. participant that is the object in the action and may or may not undergo change, e.g., beer in “Billy likes the beer’). Another useful pair of roles is that of origin (i.e., the source of the process or where the process starts, which is present at the beginning, but that does not necessarily participate throughout the whole process, e.g., “submarine” in “the missile was launched from a submarine”) and destination (i.e., the target or goal of the process). SUMO also includes the roles of instrument (i.e., tool used by an agent to bring about a process and that is not changed by the process), resource (i.e., participant that is present at the beginning of the process and that is used and consequently changed by the process, e.g., soap in a “washing hands” process), result (i.e., product of the process), among others.

BFO establishes that occurrents depend on continuants as their bearers, while continuants are subject to constant change. More specifically, it commits to the participation of independent continuants (i.e., the substantials) in occurrents (i.e., the processuals), excluding the possibility of participation of qualities and other dependent continuants. BFO also defines some dimensions to characterize a participation – active/passive, (i.e., how much agentive is the participation), directed/mediated, benefactive/malefactive (i.e., collaborating to the existence or extinction of the subject), complete/partial (i.e., considering whether the whole subject or only part of it participates as well as whether the subject is involved in the whole action or only in part of it).

Based on such dimensions, BFO brings some specific forms of participation. The first one is perpetration, i.e., direct and agentive participation of the object that perpetrates an action, which can be further specialized in initiation, i.e., initiating an occurrent, perpetuation, i.e., sustaining an occurrent, and termination, i.e., terminating an occurrent. There is also influence, i.e., having an effect on an occurrent, that is divided into facilitation, i.e., having a positive effect on the unfolding of an occurrent, and hindrance, i.e., having a negative effect on the unfolding of an occurrent. Finally, there is mediation, i.e., having an indirect role in the unfolding of an occurrent which relates other participants, and patiency, i.e., being acted upon by an occurrent.

Besides the relation of participation (between objects and occurrents), BFO also offers the converse relation of involvement (e.g., as people participate in meetings, so meetings involve people), that somewhat represents the effects of an occurrent over its participants. An involvement may come in different forms, such as creation, i.e., bringing a participant into existence, sustaining in being, i.e., sustains a participant in being, degradation, i.e., having a negative effect over a participant, destruction, i.e., put a participant out of existence, demarcation, i.e., creating boundaries for a participant, and blurring, i.e., destroying the boundaries of a participant. Finally, although dependent continuants cannot participate in occurrents and neither occurrents can involve dependent continuants, BFO offers the relations of realization (which is analogous to participation) and affecting (which is analogous to involvement), both of which having some specializations.

4.2. Summarizing the ideas

As in the previous sections, we can also divide the aspects related to participation in the three groups proposed in the introduction – Pervasive Aspects, Complementary Aspects, and Conflicting Aspects.

In light of what was considered so far in this section, we can observe that all the considered ontologies commit to some notion of participation of continuants in occurrents. Moreover, all of them commit to the idea that participants may play different roles in the context of the occurrent in which they participate (except for VEL which does not explicitly commit to the idea of occurrent-bounded roles, but neither conflicts with it). Thus, these sum up the pervasive aspects concerning participation.

Complementing this initial account, some ontologies go on further specifying these notions (summing up the complementary aspects of participation). A participation may be classified by comparing its temporal extension with that of the occurrent or with that of the existence of the continuant (e.g., constant participation, temporary participation). Participations may also be classified by whether or not the participant participates on every non-temporal part of the occurrent (e.g., total participation), by the contribution of the participant to the occurrent (e.g., perpetration, initiation, perpetuation, termination), its effect over the unfolding of the occurrent (e.g., influence, facilitation, hindrance) among other criteria. Similarly, some types of occurrent-bounded roles are proposed (e.g., agent, patient, agent-patient, origin, destination, resource, instrument, result) based on a variety of aspects. Moreover, besides its typical purpose, the notion of participation may also have further implications on other issues. For example, it may be employed as criterion for occurrent partitioning, giving rise to specific types of occurrent, such as participation in UFO, and processual role in GFO, or as unity criterion for objects, such as in YAMATO. Additionally, we can observe that there are other possible relations between continuants and occurrents (e.g., involvement, realization, affecting, being sustained by).

Nevertheless, in spite of this initial convergence, there are incompatible aspects in the different definitions of participation (in other words, the conflicting aspects regarding this facet). Some ontologies, e.g., YAMATO, define participation as an explicit requirement to the happening of occurrents, while others are not so strict about this aspect, e.g., SUMO, which states that processes may have object participants, though it is not clearly explained why it is the case or how it would happen. Additionally, there are two views regarding the sorts of continuants that can participate in occurrents. The most prevalent one restricts the possible participants to the category of existentially independent continuants, i.e., objects. This is the view adopted by, for example, BFO, SUMO, VEL, and YAMATO. However, there is also a broader view, to which GFO commits, that accepts that any type of continuant can participate in occurrents, including both existentially dependent and independent ones. Finally, some ontologies admit that not only continuants can participate in occurrents. This seems to be the case of GFO, which accepts processes to be participants of other processes through the relation of particproc (though it is not so clear how it would differ from a parthood relation). Another ontology that does not restrict participation to continuants is SEM, which accepts that not only continuants, but also places, times, and even other occurrents can play occurrent-bounded roles. Considering that playing an occurrent-bounded role implies participating in the occurrent, this definitely conflicts with the predominant view that only continuants can participate in occurrents.

5.1. Partitioning according to different ontologies

Except for ESO, all the considered ontologies provide means to represent the partitioning of occurrents, ranging from very simple approaches to richly axiomatized theories. With SEM, defining an occurrent as part of another one is simply a matter of using the hasSubEvent and subEventOf relations provided by the ontology, although it is not exactly defined which kind of part this relations yield.

The ontology underlying VEL does not explicitly define in which ways an event can be partitioned or considered part of a larger event. However, this ontology seems to implicitly commit to a certain version of mereology. First, an occurrent is described using the notion of episode (i.e., a sequence of states that obtain at each time point in a given time interval), representing the successive states the occurrent brings about during its occurrence. On top of that, VEL defines the category of unitary/solid events as having the property that they never contain two episodes such that one is a proper sub-episode of the other. Thus, representing parts of an event seems a matter of selecting those sub-episodes that represent the desired partial events. For instance, we can consider an event of making bread from its raw ingredients, represented by an episode whose sequence of states is [ s 1 , … , s n ]. Thus, we may break this episode in one such that the sequence of states is [ s 1 , … , s m ], representing the event of mixing the bread ingredients into raw bread dough, and another [ s m + 1 , … , s n ] representing the baking of bread dough into bread.

KAN defines both instance-level and type-level part-whole relations for occurrents. In instance level, an occurrent e 1 is part-of an occurrent e 2 if e 2 both temporally and spatially includes e 1 (i.e., if e 1 is constrained to the spatiotemporal region of e 2 ). By its turn, for type-level part-of relation, it is defined that an occurrent type E 1 is part-of an occurrent type E 2 if for every instance e ″ of E 2 there is an instance e ′ of E 1 such that e ′ is part-of e ″ . Besides these relations, KAN also establishes that an occurrent type may define two kinds of partial occurrent – essential events and non-essential events (with event understood here as equivalent to our notion of occurrent). Essential events are those parts that always happen given the occurrence of the main occurrent (e.g., the first half of a soccer match is an essential event of a soccer match), whereas non-essential events are those that may or may not occur (e.g., the scoring of a goal is a non-essential event of a soccer match). Considering these definitions, essential events can be defined using type-level part-of relations, while non-essential events are defined using instance-level part-of relations.

In the Event-Model-F, part-whole relations can be depicted using the mereology pattern. With this pattern, mereology is represented resorting to a situation (F:EventCompositeSituation) that includes one occurrent representing the whole (classified by the metatype F:Composite) and at least another one (classified by the metatype F:Component) representing a part of the first occurrent. In this manner, one can select two or more occurrent concepts from a domain ontology and classify one of them as the composite and the other(s) as the component(s), stating that instances of the composite have instances of the component(s) as their parts. The pattern also allows defining constraints over the time interval in which the component events must occur, the spatial region in which its participants must be located, or the spatiotemporal region of the components. With that, more than simply saying that a type of occurrent is composed of instances of some other type, it is possible to refine the conditions that an occurrent must fulfill to be considered a part of another one (e.g., we could set as parts of an epidemic its waves that spread within a radius of 100 km during the first 2 months after its beginning).

SUMO commits to temporal partition of occurrents, defining the notion of subprocess as a “temporally distinguished part (proper or not) of a process” (with process as the equivalent for our notion of occurrent). In order to represent such notion of parthood, SUMO provides the instance-level relation of subProcess, defined as a partial ordering relation (i.e., a binary, reflexive, antisymmetric, transitive relation). Besides this relation, SUMO also offers the type-level binary relations of typicalTemporalPart and typicallyContainsTemporalPart. With the former we can state that instances of a type are typically part of instances of another type, whereas the later allows us to say that instances of a type typically have instances of another type as their parts.

According to DOLCE, occurrents can have both temporal and spatial parts (e.g., a party in a two-story building may be divided into the part that is happening upstairs and the one occurring downstairs). This ontology defines an atemporal mereological relation for occurrents (i.e., a non time-indexed part-of relation). Likewise, BFO provides a temporalPart relation to account for the mereology of occurrents. Along with that, BFO defines a “temporal part” of an occurrent as the sum of all cotemporal parts (i.e., having the same temporal location) of such occurrent that are located within a given time interval – proceeding to define a “temporal slice” as an instantaneous temporal part. Every temporal part of an occurrent is also an occurrent itself. The notion of mereology to which BFO commits provides for both self-connected occurrents (represented by the concept of process, whose instances have no temporal or spatiotemporal gaps inside them) as well as disconnected wholes (represented by the concept of aggregate of processes, e.g., all the editions of the Olympic Games). Since a process has no discontinuities, all of its proper parts on the same level of granularity as the whole itself must be defined by some arbitrary division, being always fiat parts – although it is possible to find bona fide parts of a process when considering it in finer levels of granularity (e.g., the movement of individual legs in a walking process). Conversely, an aggregate of processes in a given level of granularity may be considered as a genuine process in coarser levels of granularity (e.g., a rainstorm is the aggregate of the movements of individual droplets).

For YAMATO, there are two kinds of occurrent parts: temporal parts, which result from temporal interval-based segmentation (e.g., the first half of a battle) and causal parts, which contribute to the occurrence of the overall event (e.g., a step of a walk). Any causal part is a temporal part, but not vice versa. Given that processes are simply ongoing stuff, having no unity, they can have no temporal parts, since cutting a process would make it a delimited portion of a process, which is an event. An exception can be made for processes whose time windows are not instantaneous, such that we can identify temporal parts of their occurrence patterns (e.g., a leg movement as a temporal part of a running process). For events (i.e., discrete, non-dissective occurrents, with a definite extension), there are both temporal parts (e.g., the first 30 seconds of a rock song) and causal ones (e.g., the guitar solo within the song). Besides that, YAMATO also states that events are constituted by processes (as mentioned in Section 3).

In UFO, occurrents may have other occurrents as their temporal parts, which is represented by a strict partial order relation of part-of. In that respect, UFO commits to atomistic mereology (i.e., there are atomic/non-decomposable occurrents), minimum mereology (i.e., if an occurrent is complex, it must have at least two disjoint parts) and extensional mereology (i.e., two occurrents are the same if they are composed of the same parts). Besides temporal parts, UFO also acknowledges that a complex event can be partitioned in the portions of it that are existentially dependent on each of its participants – originating participations (presented in Section 4). Since spatial qualities of events are given by its participants, partitioning an event in spatial parts seems to be encompassed by the partitioning in participations (e.g., given an orchestra concert, taking the left-side part of the concert is a matter of selecting the participations of all participants positioned on the left side of the stage). Finally, given the concepts of scene and event priorly presented, it would arguably be possible to state that events are, in some sense, parts of a scene. In other words, if a scene is whatever happens in a given spatiotemporal region, any event that happens in such a region would be part of the scene. Following, since an event is whatever happens to a selected set of focal individual qualities, such focus would be a criterion to identify the parts of a scene (to decompose a scene into its “event parts”).

Finally, GFO provides the procpart relation, that corresponds to the partition of a process into its temporal parts – so that if “x procpart y”, then the temporal extension of x is a temporal part of the temporal extension of y. Besides that, GFO may also be considered to accept the partition of a process into layer-parts – i.e., the parts of the process that refers to the participation of each of its persistant participants, giving rise to processual roles (already presented). In any case, parts of processes are always processes themselves.

5.2. Summarizing the ideas

Once more, we divide the aspects related to mereology into Pervasive Aspects, Complementary Aspects, and Conflicting Aspects, as proposed in introduction. The idea that occurrents may have temporal parts is apparently undisputed among the considered ontologies since most of them explicitly commit to such idea and the only one that does not deal with this aspect (i.e., ESO) does not pose any objection to it either. Such parts would, then, be both temporally and spatially included within the composite occurrent. It also seems uncontested that the part-of relation is a partial order relation – though it is not agreed upon whether it is a strict partial order (as in UFO), or non-strict one (as in SUMO, which does not restrict parts of an occurrent to its proper parts). These would be the pervasive aspects regarding mereology.

Enriching the picture, we can identify several complementary aspects. Some ontologies bring additional descriptions of what would be a temporal part or how the partition of occurrents in temporal parts should be regarded. From that, we end up with the idea that temporal parts are the sum of all cotemporal parts of an occurrent within a time interval (seemingly encompassing the corresponding temporal portions of any other orthogonal part of the occurrent, e.g., spatial parts, participations). As well, we should consider the existence of indivisible parts of occurrents (atomistic mereology), the notion of a complex occurrent as that composed of two or more disjoint parts (minimum mereology), and that parts of an occurrent are related to its identity (extensional mereology).

Still, probably more remarkable than that, we have the complementary features that the ontologies provide for the analysis and representation of occurrents. Besides the partition of an occurrent according to time intervals, the ontologies provide diverse criteria according to which one can split occurrents, each of them issuing a different type of part. Following the very idea of partitioning occurrents into temporal parts, we can verify whether such parts are founded on a real discontinuity on the flow of the occurrent or on an arbitrary choice of a time instant, yielding, respectively, bona fide or fiat temporal parts. Analogously to the division according to time intervals, one can partition an occurrent in spatial parts if considering the spatial region where it occurs. Paying attention to how the parts of an occurrent contribute to its unfolding, we can decompose occurrents in their causal (or functional) parts. If we turn our focus to the participants of an occurrent, we can divide it into the portions that exclusively refer to each of its participants, issuing the notion of participations (from which we may derive the notion of spatial part, if considering that all spatial qualities of occurrents are given by their participants). We can also target individual qualities that are involved in an occurrent (instead of whole participants) to split it into parts, what is used to identify events within scenes, in the UFO sense, but that could be applied to different scenarios.

Along with the idea of composition, we are also acquainted with the possibility of some occurrents being constituted by occurrents of a slightly different nature, which is proposed in YAMATO with events being constituted by processes. Here we have a somewhat restrictive notion, whose application would involve accepting the existence of a type of occurrent that conflicts with most ontologies (i.e., processes, as patterns of occurrence that can change in time). Nevertheless, the idea that an occurrent can be regarded as an additional layer of arrangements and restrictions over a simpler sort of occurrent may be a powerful tool for the modeler in need of making sense of a domain rich in occurrents.

Shifting from the instance level to the type level, we find the notion of typicality of parts of occurrents. With that, we can describe a given type of occurrent in terms of the types of occurrent whose instances usually compose the instances of the type considered. The other way around, we can also describe a type of occurrent as gathering instances that are usually part of another type of occurrent. If we choose a stricter condition, stating that the instances of a given type are always part of instances of another type, we arrive on the notion of essential part for an occurrent type. Both notions of typical and essential parts are expressed by type-level relations. Complementing these type-level concerns, we are introduced to the possibility of further constraining the possible parts of instance, not only defining the type of occurrent into which the part must be classified, but also some specific time or space regions in which the part must be included or some spatiotemporal region through which the part of the occurrent must unfold.

As a last thought, the selected ontologies present some possibly conflicting views on partitioning. For example, depending on the view adopted, one can acknowledge the existence of instantaneous temporal parts of prolonged occurrents (such as in BFO) or must reject it, having only prolonged occurrents as parts of prolonged occurrents (as in GFO). We can also mention the view according to which it is possible to have an occurrent composed of parts that are not spatiotemporally connected (as in the case of aggregate of processes in BFO, that accepts instances such as the set of all wars during a period of time), which is not accepted (at least explicitly) by the other ontologies – specially by GFO view, that requires that the boundaries of an occurrent (i.e., the temporal slices of the continuants participating on the occurrent) are ontically and causally connected.

6.1. Causation according to different ontologies

The notion of causation is found in most of the ontologies analyzed, except for SEM, ESO, and VEL. In YAMATO, the idea of causation is not clearly defined, being diffuse on its use for describing different aspects of the ontology. It is present in the link between the internal processes of an object (i.e., those that sustain the object in being) and its external processes (i.e., those that are enacted by the object, and that determine object’s nature and identity) – with the former causally contributing to the later. It is also employed to convey the notion of causal parts of an occurrent (i.e., those parts that causally contribute to the occurrence of the overall occurrent). Besides those rather structural concerns, YAMATO also acknowledges causation as an aspect of the unfolding of occurrents, stating that, in a causal process, each successive phase is caused by the earlier phases. Additionally, this aspect is used to set apart processes (i.e., which have such causal link between their phases) from pseudo-processes (i.e., in which there is no causal link, being its successive phases independently caused by some external process, e.g., the motion of the cursor on a screen, or the movement of a shadow).

SUMO conveys the notion of causation by means of the relation causes between occurrents, with p 1 causes p 2 meaning that p 1 brings about p 2 (where both p 1 and p 2 are instances of some type of occurrent). It is a binary, asymmetric relation. The causes relation is a specialization of relatedEvent, which is defined as a relation between separate individual occurrents – not between an occurrent and its parts – so that it cannot accept causation from parts to their wholes. SUMO also provides the causesSubclass relation to indicate a causation relation between two types of occurrent. Then, if P 1 causesSubclass P 2 , for every occurrence of an instance p ″ of the occurrent type P 2 there is an occurrence of an instance p ′ of the occurrent type P 1 such that p ′ causes p ″ (e.g., Killing and Death).

In KAN, causal relation is regarded as a complex binary relation that can be established between continuants, occurrents, from continuants to occurrents (e.g., if John sets fire to a house, the cause of the fire is John), and vice-versa, although only causation between occurrents is further detailed. Causation between occurrents is represented both in instance and type levels. In instance level we have eventInstanceCausalRelation, a transitive relation expressing the general notion of causation – i.e., e 1 causes e 2 means that the occurrent instance e 1 causes the occurrent instance e 2 . According to the temporal relation between the relata, this instance-level relation is further specialized into disjointCausalRelation (i.e., when there is a temporal interval between the cause and effect occurrents), continuousCausalRelation (i.e., when the effect occurrent happens right after the cause occurrent, and then both events temporally meet in the sense of Allen’s relations (Allen, 1983)), overlappingCausalRelation (i.e., there is an overlapping between cause and effect occurrents, with a time interval during which both occurrents are taking place), and partialCausalRelation (i.e., the temporal interval during which the effect occurrent happens is included in the temporal interval of the cause occurrent). Following from that, the intransitive type level causal relation eventClassCausalRelation conveys the idea that any time a type of occurrent happens, an instance of a second type of occurrent will also take place (i.e., if occurrent type E 1 causes occurrent type E 2 , for every instance e ′ of E 1 there is an instance e ″ of E 2 such that e ′ causes e ″ ).

In Event-Model-F, causation is a unidirectional relation, so that causes happen before effects and are necessary conditions for them. Moreover, a causation relation is justified by some (perhaps implicit) underlying causal theory. Cause and effect are roles that only occurrents can play, so that continuants that might be regarded as cause or effect are in fact participants of occurrents playing such roles. There is only one (maximal) cause for an effect (i.e., the mereological sum of all concurrent causes) and only one (maximal) effect for a cause (i.e., the mereological sum of all effects). In order to account for this definition of causation, the model provides a causality pattern, in which causation is represented by a causality situation (F:EventCausalitySituation), that gathers two types of occurrent (classified as F:Cause and F:Effect), with their respective participants indicated by the already mentioned participation pattern. It also includes a DUL:Description classified as F:Justification (i.e., the justification/theory that justifies the causal relation presented using the pattern). If there are multiple causes (or effects) in the relation, they are gathered into a single aggregated cause (or effect) using the mereology pattern. This pattern can also be specialized to represent (domain) specific kinds of causal relationships. Besides causation, Event-Model-F also provides the correlation pattern, allowing representing correlated occurrents, understood as occurrents that have a common cause.

Except for the concept of scene, in UFO occurrents are regarded as transitions from a situation to another (with the concept of situation understood as a particular configuration of a part of reality which can be understood as a whole). Thus, for each occurrent of such kind, there is a pre-state situation, which obtains at the initial time instant of the occurrent. Such pre-state exemplifies a state of the world that satisfies all the necessary and sufficient conditions for the manifestation of the occurrent, triggering it – so that there is a unique situation that triggers a particular occurrent happening. Likewise, there is a pos-state situation, which is brought about by the occurrent, obtaining at the ending time instant of the occurrent. Again, there is a unique maximal situation brought about by an occurrent, encompassing all the changes it causes in the world. On top of this framework, UFO defines the instance level relation of directlyCauses, with e 1 directlyCauses e 2 meaning that the occurrent instance e 1 brings about a situation that triggers e 2 . It is a strict partial order relation and implicates that e 1 meets e 2 . From that, UFO defines the strict partial order relation causes, with e 1 causes e 2 meaning that either e 1 directly causes e 2 or e 1 causes an intermediary event that causes e 2 , implicating that e 1 happens before e 2 .

Yet not providing a detailed treatment, BFO acknowledges both successive causality (i.e., causal triggering, with an occurrent causing another, later occurrent that is the realization of a disposition) and simultaneous causality (i.e., causal dependence, with both occurrents happening simultaneously, causally contributing to each other, e.g., according to Boyle’s law, a rise in the temperature of an amount of enclosed gas causes the simultaneous increase of its pressure, and vice-versa).

DOLCE only deals with causal relations involving occurrents of a very simple sort, namely, basic quality changes (i.e., changes of a single continuant along a single quality type, e.g., a color change) – not covering (but also not denying) causation between complex occurrents (i.e., changes involving multiple aspects and/or multiple continuants). Moreover, its notion of causation (in the sense adopted in this paper) relies on three main constraints (i.e., structural, causality and circumstantial constraints), which act as dependences between basic quality changes that capture orthogonal aspects of this notion. As a last note, DOLCE makes a distinction between the notions of causality (i.e., law-like relation between types of occurrents) and causation (i.e., actual relation that holds between individual occurrent instances).

GFO also acknowledges a relation between causes and their effects, supporting regularity – i.e., causation as invariable patterns of succession (Hitchcock, 2018) –, counterfactual dependency – i.e., if A had not occurred, C would not have occurred (Menzies, 2017) –, and manipulability – i.e., causes regarded as handles for manipulating effects (Woodward, 2016). However, it adopts a view that diverges from most the other ontologies, defining causality as a relation between continuants (i.e., “special relation between presentials”), so that it would be continuants to play the roles of cause and effect rather than occurrents. In spite of this view of continuants as causal relata, the discussion in Herre (2010) suggests the possibility of extending the definition of causal relation to cover occurrents as causal relata as well. In fact, this idea is implicitly present when it is stated that ontical relations connect both continuants (presentials) and occurrents (processes), due to the fact that ontical relations are regarded as integrated systems of spatiotemporal and causal relations – in such a way that being ontically connected is being in some way causally connected as well.

6.2. Summarizing the ideas

As pervasive aspects of causation, the considered ontologies seem to agree on a general notion of causation as a binary relation holding between occurrents that represent a cause and an effect, with causes happening before effects. Moreover, many of the ontologies appear to understand causation in a sense closer to that of counterfactual dependence (e.g., Event-Model-F, with causes as necessary conditions for effects). This provides an initial approach to include causation concerns on some model.

On top of this initial basic notion of causation, the different ontologies add further specificities and unveil diverse complementary aspects concerning the definition of this relation. It is accepted the existence of versions of this relation which hold between individual instances – so as to state that a particular occurrence causes another particular occurrence – and others that are established between types – defining that the occurrence of an instance of a given type would cause (or be caused by) the occurrence of an instance of another type. It is also proposed the notion of situation as something close to the idea of causal link, with a cause occurrent eventually bringing about a situation that triggers the occurrence of an effect occurrent. Along with the idea of causation as a binary relation, it is also present the view that each cause causes a single effect and that each effect is caused by a single cause. With that, a single cause would be represented either by the mereological sum of all the occurrents that makes up the single cause – or by the maximal situation that triggers an effect occurrent. Likewise, a single effect would be the mereological sum of all occurrents that are triggered by a single cause – or the maximal situation brought about by the single cause occurrent, encompassing all the changes it performed). Finishing the list, besides counterfactual dependence, causation can be also regarded as regularity and manipulability.

Causal relation between instances is generally regarded as transitive – although this is not necessarily the case for type-level causation. The ontologies also present some specializations of this relation, according to the temporal relation between causes and effects (as in KAN, yielding disjoint, continuous, overlapping and partial causal relations). Causal relation is also specialized according to the existence or not of intermediary occurrents between a cause and its final effect (as in UFO, which specifies the directly-causes relation). Besides that, it is also considered the possibility of specializing the relation of causation to domain-specific types of causation. Along with that, a justification according to some underlying causal theory is also considered as part of the description of a causal relation.

The ontologies also present some seemingly irreconcilable divergences (i.e., conflicting aspects). Each of the ontologies accepts (at least implicitly) the notion of causation between occurrents. However, whereas some ontologies explicitly restrict this relation to hold solely between occurrents, others (i.e., KAN, GFO) accept that continuants may also be causal relata. Furthermore, while most ontologies regard causation as relating only things of the same nature (i.e., occurrent to occurrent, continuant to continuant), KAN acknowledges any combination (i.e., including the possibilities of continuants causing occurrents and vice-versa). Another source of conflict is the direction of the relation. Again, all the ontologies accept the unidirectional causation (from cause to effect, with the former preceding the later). Nevertheless, while some ontologies only accept this kind of relation, others consider the possibility of bidirectional/simultaneous causation (with both occurrents causally contributing to each other). As a last divergent aspect, SUMO explicitly states that causal relation can be only established between separate individual occurrents (not between an occurrent and its parts), while YAMATO implicitly commits to some causal relation between an occurrent and its causal (or functional) parts.

Finally, the ontologies provide some cases in which the establishment of causal relations can be witnessed. The already mentioned link between an occurrent and its causal parts is one of such cases. Another one is related to the particular view of YAMATO on the nature of objects as interfaces between internal and external processes. Causation is also present in the notion of correlation (present in the correlation pattern in Event-Model-F and in the notion of pseudo-process in YAMATO). The last, and probably the most significant one, is the causal relation between the successive stages of an occurrent, be them either the temporal parts of the occurrent or the temporal slices of continuants exhibited at each time instant (as in GFO). In the case of GFO, it is also present on the inertial principle, according to which there must be some external cause for a state change (with state considered as a subtype of process).

7.1. Observer-related issues

At first glance, anything related to the idea of observer seems to better fall under epistemological matters rather than ontological ones. It is certainly the case when we care about how we acquire knowledge and update our beliefs – or, to put it simply, when we are on the position of an observer trying to make sure that what we know is correct. However, we may think about it from an external standing point, as the dynamics involving such a thing as an observer and things such as ideas, views, or beliefs which the observer gets upon the observation of something else. Put this way, talking about the existence of such things, ontological considerations become appropriate.

In part, this issue emerges from the problem of reconciling conflicting views about any given single instance of occurrent – and some ontologies provide means to meet this need. For instance, as stated in van Hage et al. (2011), SEM is “meant to record different, possibly conflicting, descriptions of an event as co-existing facts”. For that purpose, SEM includes the notion of view as a property that only holds according to a certain authority, which may or may not participate in the event. Views can be used to constrain roles, that would apply for continuants, places, times, and even for other occurrents. Therefore, using views, one could represent, for example, different roles for participants (e.g., hero, villain) or occurrents (e.g., victory or defeat, improvement or worsening), according to different authorities. With a similar purpose, Event-Model-F offers the interpretation pattern to represent divergent interpretations over the same occurrent. In this pattern, an interpretation is a situation that includes an occurrent (the F:Interpretant) and another situation representing the context according to which the occurrent may be interpreted. This contextual situation can be of any of the types represented by the other patterns the model provides (e.g., with causality situation, one can represent an interpretation regarding the cause or effect of an occurrent; for interpretations with respect to the participants, one can use a participation situation).

On the other side, observer-related issues are also on the very selection of what we are talking about when referring to occurrents. This faintly, implicitly appears in KAN when distinguishing object events and object states (i.e., occurrents which have a recognizable main participant) from environment events and environment states (i.e., those described using solely location and time, without mentioning any participant). Rather than solely a matter of observation, it seems to reveal a more fundamental feature, that allows classifying occurrents into two types: those that allow someone to spot the main participant and those that do not. In other words, some events are thought of as having a focal point whereas others are more diffusely described. Further in this line of thought, a follow-up work related to UFO (Guarino & Guizzardi, 2016) explicitly contemplates the issue of focus for events. But, instead of considering focal participants, it explicitly brings the finer notion of focus as a collection of individual qualities (i.e., focal qualities) to whose behavior one pays attention when identifying and individuating occurrents as well as describing them. Then, focus is used to divide occurrents into those that present a focus (i.e., events) and those that do not (i.e., scenes).

In somewhat different ways, both ontologies acknowledge the role of a focal point (or its absence) on classifying occurrents, making a distinction between diffuse and focal occurrents. For modelers, it also gives some aid in how to recognizing occurrents (i.e., how the abstraction of an occurrent could be conducted): either by gathering all the happenings in some spatiotemporal region/environment or by paying attention to to some specific portion of it (delimited by selected focal qualities or a focal participant). Moreover, the finer notion of focus as a set of focal qualities can help to solve some problems concerning individuation of occurrents – for example, as demonstrated in Guarino & Guizzardi (2016), it can solve the problem of individuating the events that are happening when a sphere is simultaneously heating and spinning (Davidson, 1969).

7.2. Roles for occurrents

It seems well accepted the idea that continuants may play roles, that come in two main types: relational roles (i.e., acquired in virtue of participation in some relation, e.g., the president of a council) and occurrent-bounded roles (i.e., acquired in virtue of participating and behaving in certain way with respect to an occurrent, e.g., the agent of an occurrent). Both types seem to present a relational nature (i.e., something plays a role with respect to something else, that is either an occurrent or another continuant) and a contingent nature (e.g., the role holder can play the role for just a period during its life).

On the other hand, the idea of occurrents playing roles of any kind usually faces objections, mainly due to the contingent nature of such idea, that seems to conflict with the widespread idea that events cannot suffer change – so that, if it cannot change, how could it acquire and lose roles?

In spite of that, among the considered ontologies, there are suggestions about the idea of occurrents playing roles. In the paper presenting the Event-Model-F (Scherp, Franz, Saathoff & Staab, 2012) it is stated that “only events can play the roles of cause and effect”, which certainly has a relational nature (i.e., an occurrent only plays the role of cause in relation to some other occurrent – the effect). Additionally, in some sense not explicit in the ontology, the role of cause could also be considered contingent (since, in the example of cause and effect, an occurrent is not a cause until its effect occurs). Analogous reasoning can be developed with the roles presented on other patterns brought by the ontology (e.g., events play the role of components in relation to a composite solely when the composite event is completed, with all its parts; interpretants play such role in relation to an interpreter just during interpretation process). Furthermore, whereas the notion of role in Event-Model-F is implicit, SEM explicitly allows specifying that an occurrent plays some role in the scope of another occurrent, although there is no further discussion on why considering this possibility or how such role would exist, since it would apparently imply the participation of an occurrent in another occurrent, which seems an odd idea.

The possibility of occurrents playing roles is certainly useful, both for allowing the representation of cases otherwise not possible and for helping to make explicit on models the contingent and relational nature of some types of occurrent. Along with that, accepting this option suggests diverse ways to analyze the context in which an occurrent happens. It may lead the modeler to take a step backward to look for other entities whose relations are necessary to make clear the nature of some occurrent. Also, considering this possibility may lead the modeler to search for some other occurrent that may serve as context, in which scope the role is played.

7.3. Contextual occurrents

Some ontologies suggest that, besides being what happens with their participants, some occurrents may be regarded as frames for the happening of other occurrents. The simplest case is that of scenes in UFO, defined as everything happening in a suitably restricted spatiotemporal area and from which events emerge – so that we can say that any given event happens in the context of a scene. BFO presents the concept of setting (or processual context) as an occurrent which stands as a spatiotemporal surrounding environment (or four-dimensional site) for other occurrents, in which they can be located to occur (e.g., a clinical trial can be the processual context for the discovery of a new treatment). In a similar approach, GFO presents situoids, which are occurrents representing an integrated whole made up of other occurrents.

Weaker notions of contextual occurrents can be found in YAMATO and KAN. YAMATO presents the concept of situation as an occurrent representing a state of affairs or the interaction pattern among participants (and their external states). It is introduced as a state shared by many participants, in opposition to the concept of state (which implies that the state is possessed by a single entity). Thus, in some sense, a situation could be seen as the sum of other smaller situations and/or individual states, being the context in which such smaller occurrents happen. Environment states in KAN, representing a state being kept by no particular identifiable object, may be regarded in a similar way, as the context for the states of possibly unfocused objects participating in the occurrent.

There are also restricted versions of this notion, applied to single entities, called histories in BFO and GFO or life in DOLCE. Generally speaking, they represent the mereological sum of all occurrents exclusively dependent on the continuant of interest or those that take place on the spatiotemporal region occupied by the continuant. This way, it would make sense to say that some occurrent took place in the context of the history/life of a particular continuant.

This notion of contextual occurrents provides an additional way to look for occurrents. Besides that of processes (in the sense of “cyclic occurrents”, e.g., walking), events (as transitions between situations, e.g., the baking of bread) and states (as some coherence that is actively kept by participants, e.g., being seated), one can take a step backwards and look at the big picture, identifying complex occurrents that frame others. Such a view would help to make explicit some diffuse, but yet relevant, occurrents (e.g., how could the Cold War occurrent be considered without the notion of contextual occurrent?). Furthermore, it may also be useful when it is necessary to capture faint or diffuse connections between occurrents, that would not be possible to depict using more direct relations, such as those of causation, correlation, or composition (e.g., all the sales that happen in a trade fair).

8.1. Participant-related criteria

The most numerous occurrent classification criteria are those related to the participants of the occurrent. The simplest criteria is the number of participants involved (i.e., if the occurrent involves a single or multiple participants), that is used to differentiate simple process from complex process in GFO, single-actor action from multi-actor action in YAMATO, group action from intransitive action and transitive action in KAN, and that, in a follow-up work related to UFO (Guarino & Guizzardi, 2016), is behind the definition of relational event as opposed to (not explicitly mentioned) single-participant events.

Matters related to the agent role also serve as occurrent classification criteria. Agentivity (i.e., whether or not there is some participant playing the role of agent, regardless its intentionality, e.g., the fire of a stove burner boiling some water), is used in the division of action_behavior_motion vs. phenomena in YAMATO. Further on, intentionality (i.e., whether or not the agent of the occurrent deliberates its action, e.g., a person waving to another) is another criteria, used to distinguish action from behavior in YAMATO, activity from phenomenon in DOLCE, natural event from artificial event in KAN, and, in some measure, action event from non-action event in UFO. KAN also employs agent-patient superposition criteria (i.e., whether or not the participants simultaneously play the roles of agent and patient) to distinguish intransitive action from transitive action.

Structural concerns are also taken into account. Internality (i.e., whether the occurrent affects the internal or external qualities/state of the participant) is used to distinguish states (either dynamic or static) from events (either natural or artificial) in KAN and internal states from external states in YAMATO. Materiality (i.e., whether or not participants present material structure) is used in GFO to divide material process and quality process. YAMATO uses ownership (i.e., whether or not the occurrent is possessed – in a loose sense – by a single entity) to differentiate state from situation and goes further with holder type (i.e., the type of the entity that owns the state), to distinguish object state from process state.

What happens with the participants is also considered. YAMATO distinguishes action_behavior from motion, which seems to raise location change as an occurrent classification criteria. UFO employs existence status (i.e., the effect over the existence of participants) to distinguish creation, change and destruction occurrents. Finally, in a broader approach, KAN orthogonally classifies occurrents according to their semantic function (i.e., the type of effect the occurrent has over the participants). With that, KAN distinguishes state change (i.e., change on the qualities of the participant), temporal existence change (i.e., the participant exists in a time point, but not on the next/previous instant), spatial existence change (i.e., movement), cardinality change (i.e., change on the number of continuants), object identification change (i.e., a change in the essential qualities of the participant, so that it can no longer be recognized as the same individual), and comparison (i.e., observing a change in some quality of the participant).

8.2. Part-whole-related criteria

Part-whole issues are also a rich source of occurrent classification criteria. Atomicity/Divisibility (i.e., whether or not the occurrent can be broken into proper parts) is used, with respect to temporal parts, in UFO to distinguish complex events from atomic events, in DOLCE to distinguish accomplishments (which have parts) from achievements (no parts) and in VEL to distinguish unitary/solid event (i.e., those with no proper part) from the others (that have parts, but that are not explicitly mentioned as a type). YAMATO also use this criterion to the divisions of composite action vs. simple action and composite behavior vs. simple behavior, which seems to also comprise the notion of functional part though. Wholeness (i.e., whether or not the occurrent represent a composite whole) is used in Event-Model-F to distinguish composite from component, and in BFO to distinguish process from fiat process part and process boundary. Contextual wholeness (i.e., whether or not the occurrent represent a contextual whole) differentiates configuroids from states and discrete/continuous processes in GFO. BFO employs spatiotemporal connectedness (i.e., whether or not there are spatiotemporal gaps inside the occurrent) to distinguish process aggregates (i.e., the mereological sum of different processes) from processes. YAMATO uses functional structure (i.e., whether or not the occurrent is the arrangement of functional parts) to differentiate process_1 (i.e., structured processes) from process_2 (i.e., simple sequences). This criterion seems to be also present on the distinction between extrinsic accomplishment (i.e., just a chunk of a process) from intrinsic accomplishment (i.e., an event in its own right, as a functional whole). Finally, in DOLCE, cumulativity (i.e., whether or not the mereological sum of instances of the same type is still an instance of such a type) is used to distinguish stative from event, and homeomericity (i.e., whether or not the temporal parts of an occurrent are of the same type of the whole) to distinguish states from processes.

8.3. Time-related criteria

Regarding temporal aspects, instantaneity (i.e., whether the occurrent lasts for a single instant or has a longer duration) is used to distinguish punctual event from durative event in VEL, instantaneous change from continuous change in GFO, and instantaneous event from semi-instantaneous event in YAMATO. YAMATO also uses anchoring point (i.e., the time point of reference on the temporal extension of the occurrent) to differentiate start-open event (i.e., anchored on its end point, e.g., arriving), end-open event (i.e., anchored on its start point, e.g., leaving), and both-open event (i.e., with both ends not well defined, being anchored in some other point within the occurrent, e.g., passing by). Finally, YAMATO uses relative temporal-position (i.e., the position of the occurrent in relation to other) to distinguish starting event from ending event.

8.4. Dynamics-related criteria

Regarding dynamics (here understood as related to whether occurrent is static or dynamic, and if dynamic, in which extent), the criteria of change nature (i.e., whether or not the occurrent, as a whole or any of its parts, represents some type of change) is used on the division between static event and dynamic event in ESO, static state and dynamic state in KAN, and to distinguish state from discrete process and continuous process in GFO. Discrete change (i.e., whether or not the occurrent presents an instantaneous, non-gradual change) appears in GFO as a distinction between discrete process and continuous process. Adopting another view, YAMATO considers state-basedness (i.e., whether the occurrent is related to the maintenance of a situation/pattern of occurrence or it is related to a situation transition) to distinguish stative from event.

8.5. Focus-related criteria

Considering focal and observer issues, we have focus (here broadly understood as the presence of some outstanding feature to which we direct our attention to identify the occurrent) distinguishing events from scenes in UFO and object events/states from environment events/states in KAN. There is also instant-basedness (i.e., whether the distinguishing feature of the occurrent is instantaneous or must be observed for a longer interval) that is used in YAMATO to distinguish instant event (i.e., either an instantaneous occurrent or an occurrent whose main component is an instantaneous occurrent) from ordinary event and to distinguish stative_2 (i.e., states and situations, whose configuration can be verified in any time point) from process (i.e., whose occurrence pattern is not instantaneous).

Pervasive aspects refer to reasonably settled matters – apparently found on solid philosophical arguments and compliant with ordinary experience. Thus, they can be considered a basic common view about occurrents to which the considered ontologies commit. Not all of them are mandatory properties of occurrents, but rather properties that every (or at least most) of the ontologies explicitly accept. Likewise, they are non-exclusive, so that it is possible for other ontologies to accept additional alternative views concerning such aspects (e.g., having temporal parts does not exclude the possibility of having parts according to some orthogonal criteria). Considering that, the sum of such aspects constitutes a description of occurrent that may qualify as a basic, independently endorsed definition, which should not be overlooked and that apparently yields an initial picture for the treatment of occurrents (e.g., for identification, representation, reasoning).

Complementary aspects can be regarded as additional interesting ideas that researchers developed as a result of contemplating diverse domains or in response to different modeling needs. As minority views, they might represent less prevalent aspects or ideas that were more recently brought to debate – which may offer opportunities to consider what would be the implications if such a view were adopted in a greater variety of cases. Therefore, not including them on models may not cause major problems. On the other hand, since they do not conflict with commitments of the other ontologies (especially the pervasive aspects), they can be regarded as optional tools to be freely employed to enrich the model according to the particularities of the targeted domain or task.

In addition, these aspects seem to refer to things we only perceive when getting a closer look at the occurrent, as opposed to pervasive aspects, that are apparently immediately recognizable. For example, it is clear that occurrents happen in time, but conceiving them as quality manifestations requires further analysis (e.g., we must first identify the participants in order to determine which of their qualities are being manifested). It is also relatively straightforward to identify the participants of an occurrent (since we arguably only perceive it by observing its effects over the participants), but determining their roles demands more attention. Likewise, temporal parts are evident, which is not the case for the orthogonal divisions (e.g., the various participations that compose an occurrent are frequently intricately related to each other, being hard to individuate as significant single occurrents). Still, general causation between two occurrents seems intuitive, but it requires further thought to recognize additional characteristics of this notion (e.g., transitivity, situations as causal links between occurrents).

Conflicting aspects represent irreconcilable options, yielding modeling decisions that practitioners must take when analyzing a domain. This group encompasses both acknowledged open issues (e.g., the ontological priority between continuants and occurrents) and proposals of divergent paradigms (e.g., occurrents playing occurrent-bounded roles). Whilst some of them sound as minor divergences (e.g., possibility of instantaneous parts), some others relate to deeper problems, with far-reaching implications, leading to relevant differences on the models of specific domains (e.g., ontological priority, occurrents undergoing changes or playing roles w.r.t. other occurrents, bidirectional causation, continuants as causation relata).

Regarding occurrent classification criteria, it would probably not be too far-fetched to say that they probably help to uncover what we first notice when we think about occurrents. Therefore, having an instance of occurrent, it would be the case that we first classify it into the suitable type to later complete its description with additional details. Thinking this way, we may entertain an idea about what catches our attention the most on occurrents (always keeping in mind the limitations of the sample of considered ontologies). Observing the general facet/aspect to which the criteria are related, it is noticeable that the majority of the criteria are related to participant and part-whole issues, which may suggest that these are the main targets when someone is trying to make sense of some occurrent. On the other hand, moving the attention to the specific criteria and how frequently each of them is used across the considered ontologies, the most recurrent classification criteria are those related to how may participants the occurrent involves, whether or not the occurrent is a deliberate action, how it can be partitioned, which is its duration, and in which extension it represents a change. In light of that, the aspects/facets on which these criteria are based may configure, along with the pervasive aspects, an initial target for the task of analyzing occurrents.

9.2. Impacts over the analysis and modeling of occurrents

Analogously to a painter who has more options in his color palette to better depict the scene s/he observes, having a more extensive list of aspects to bear in mind when analyzing/modeling occurrents may allow the modeler to make more precise representations. Knowing about the diverse possible aspects under which to consider an occurrent, the modeler can choose the right combination to reveal and make explicit the relevant details of the domain, making it easier to convey the conceptualization that is already shared by the community.

The sum of the pervasive aspects gives us a rich definition of occurrents: occurrents are things that are not continuants, that happen in time accumulating temporal parts, that are associated with time intervals, that have continuants as participants (which may play different occurrent-bounded roles), that may represent changes or maintenance of states, and that are the relata of causal relations. This definition allows differentiating occurrents from continuants (which may seem somewhat obvious, but that may not be immediately accessible for the modeler since this is not a native distinction in most modeling languages). On top of that, the modeler analyzing an occurrent is led to ask when it has happened, which objects participated in it (and what were their contributions), and if it can be broken into significant temporal parts – ending up with a meaningful picture of the occurrent in hand. It would raise similar questions for a modeler trying to define a concept representing a type of occurrent from the domain in hand (e.g., how long is the interval typically associated with instances of such type, which kinds of participant it has, which are the relevant parts).

Moreover, acknowledging that occurrents can represent both change and maintenance of states broadens the scope of occurrent identification. Whereas recognizing an occurrent when it represents a change in the state of the world seems straightforward (i.e., it immediately follows from the perception of the difference from one instant to another), conceiving that there may be an underlying occurrent that is “keeping things as they are” requires a richer conceptual framework in order to come up with and investigate such possibility. Finally, viewing occurrents as relata of causal relations opens for the modeler the possibility of exploring the causation chain. Instead of just attributing the happening of some occurrent to the involved participants, s/he is also led to verify which were the previous events that led to the occurrent in analysis, as well as the possible occurrents that may happen in consequence of the present occurrent.

The remaining complementary or conflicting aspects have a similar impact. Concerning mereology, the different kinds of parts (spatial, causal/functional, by participant, by focusing qualities) indicate the joints into which an occurrent may be meaningfully partitioned, beyond merely arbitrary temporal parts. In addition, accepting that an event may have temporarily-disconnected parts permits us to recognize certain temporally scattered occurrents as a single occurrent (e.g., a soccer competition that happens during a whole year, but that is composed of weekly matches) in an analogous way we identify spatially scattered occurrents (e.g., an art festival composed by simultaneous artistic interventions in different theaters).

Participation is in a comparable situation. Committing to different types of participation (e.g., with respect to temporal extension) enables defining different degrees of involvement of the participants. Additionally, committing to different specific types of occurrent-bounded roles (e.g., agent, result) permits distinguishing what is the contribution of each participant to the occurrent as a whole. A similar case can be made for alternative continuant-occurrent relations – e.g., realization, affecting, being sustained by.

Regarding the nature of the occurrents, complementary definitions may help to reveal the existence of underlying occurrents that would not be identified with a simpler view. For example, considering occurrents as manifestations of qualities of objects allows the individualization of certain superimposed events – happening at the same spatiotemporal region, involving the same objects – that otherwise could only be seen as a single occurrent (e.g., the problem of the sphere simultaneously heating and spinning, presented in Section 7.1).

Considering occurrents as transitions between situations13

Some commitments may also lead to unexpected consequences, that can be further explored by the modeler. For example, admitting that observers (which are continuants) may have different views or interpretations about a single event seems to imply the existence of a different kind of relationship between continuants and occurrents, which is not that of participation or those closely related ones (e.g., involvement, realization, being sustained by). In fact, differently from such “participation-like” relations, this “observing-like” relationship may be even established between a continuant that exists at a time and an occurrent that had happened long before the continuant came into existence (e.g., when a historian is indirectly observing a historical event by means of documents).