Intelligenza Artificiale

Accountability is often seen as a key notion in distributed systems, both in the human world and in software, on top of which interaction is built, together with the sibling notion of responsibility. However, there is limited support for the specification and use of such concepts in computational settings. This paper proposes an information model for computational accountability, which describes what data have to be available to allow, in any situation of interest arising from a group of interacting parties, the investigation and identification of accountability. The characterization of accountability provided by the model is grounded in the notions of just expectation and control. The information model is expressed in Object-Role Modeling (ORM), since it is well-suited to capture the relational nature of the accountability concept. An advantage of our model is that a designer can verify the consistency of a domain w.r.t. a set of accountability requirements. The paper exemplifies this checking by means of an Answer Set Program (ASP).

The database (DB) landscape has been significantly diversified during the last decade, resulting in the emergence of a variety of non-relational (also called NoSQL) DBs, e.g., xml and json-document DBs, key-value stores, and graph DBs. To enable access to such data, we generalize the well-known ontology-based data access (OBDA) framework so as to allow for querying arbitrary data sources using sparql. We propose an architecture for a generalized OBDA system implementing the virtual approach. Then, to investigate feasibility of OBDA over non-relational DBs, we compare an implementation of an OBDA system over MongoDB, a popular json-document DB, with a triple store.

This article is an extended and revised version of an article that appeared in the proceedings of the 17th International Conference of the Italian Association for Artificial Intelligence (AI*IA) [4].

 The search for unfamiliar experiences and novelty is one of the main drivers behind all human activities, equally important with harm avoidance and reward dependence. A recommender system personalizes suggestions to individuals to support and guide them in their exploration tasks. Personalization mechanisms and recommender systems limit serendipitous encounters by selectively guessing the next item to show to users and potentially leading them into so-called filter bubbles. In the ideal case, these recommendations, except of being accurate, should be also novel. However, up to now most platforms fail to provide both novel and accurate recommendations. For example, a well-known recommendation algorithm, such as matrix factorization (MF), tries to optimize only the accuracy criterion, while disregarding the novelty of recommended items. In order to counteract the filter bubble, we propose two models, denoted as popularity-based and distance-based NMF, that allow to trade-off the MF performance with respect to the criteria of novelty, while only minimally compromising on accuracy. Our experimental results demonstrate that we attain high accuracy by recommending also novel items.

The recent breakthroughs in the field of deep learning led to state-of-the-art results in several NLP tasks, such as Question Answering (QA). Unfortunately, the requirements of such neural QA systems are very strict due to the size of the involved training datasets. In cross-linguistic settings these requirements are not satisfied as training datasets for QA over non-English texts are often not available. This represents the major barrier for a wide-spread adoption of neural QA methods in NLP applications. In this paper, the acquisition of a large scale dataset for an open-domain factoid question answering system in Italian is discussed. It is obtained by automatic translation and linguistic elicitation of an existing English dataset, i.e. the SQuAD question-answer pair corpus. Even though the quality of the resulting corpus for Italian might not be completely satisfying, our work allowed to generate more than 60 thousand question-answer pairs. In the paper the impact of this resource on the QA process over the Italian Wikipedia is studied, according to different training conditions and architectural constraints. A comparative evaluation against the English version, in line with standards in the SQuAD literature, is carried out. The outcomes show that the results achievable for Italian are below the state-of-the-art for English, but the ability of learning not to respond (i.e. the adoption of techniques for detecting question whose answers are simply not available, i.e. EMPTY set of answers) allows the system to pursue reasonable levels of precision. This make it already usable within realistic application scenarios. Finally, an error analysis is presented that suggests possible future research directions on still critical but highly beneficial enhancements, in view of concrete QA applications in Italian.

The Operating Room Scheduling (ORS) problem is the task of assigning patients to operating rooms, taking into account different specialties, the surgery and operating room session durations, and different priorities. Given that Answer Set Programming (ASP) has been recently employed for solving real-life scheduling and planning problems, in this paper we first present an off-line solution based on ASP for solving the ORS problem. Then, we present techniques for re-scheduling on-line in case the off-line schedule can not be fully applied. Results of an experimental analysis conducted on benchmarks with realistic sizes and parameters show that ASP is a suitable solving methodology also for the ORS problem. This analysis has been performed with a web framework for managing ORS problems via ASP that allows a user to insert the main parameters of the problem, solve a specific instance, and show results graphically in real-time.

This paper investigates the role of coherence constraints in recognizing facial expressions from images and video sequences. A set of constraints are introduced to bridge a pool of Convolutional Neural Networks (CNNs) during their training stage. Constraints are inspired by practical considerations on the regularity of the temporal evolution of the predictions, and by the idea of connecting the information extracted from multiple representations. We study CNNs with the aim of building a versatile recognizer of expressions in static images that can be further applied to video sequences. First, the importance of different face parts in the recognition task is studied, considering appearance and shape-related features. Then we focus on the Semi-Supervised learning setting, exploiting video data, where only a few frames are supervised. The unsupervised portion of the training data is used to enforce three types of coherence, namely temporal coherence, coherence among the predictions on the face parts and coherence between appearance and shape-based representation. Our experimental analysis shows that coherence constraints improve the quality of the expression recognizer, thus offering a suitable basis to profitably exploit unsupervised video sequences, also in cases in which some portions of the input face are not visible.

In this work we exploit a nonmonotonic Description Logic of typicality as a tool for the generation and the exploration of novel creative concepts. Our logic, called T ^CL , is able to deal with the phenomenon of prototypical concept combination, which has been shown to be problematic to be modelled in formalisms like fuzzy logic. The logic T ^CL is the result of combining three main ingredients: (i) the logic of typicality ALC + T R , that allows one to describe typical properties of a class by means of a typicality operator, (ii) the distributed semantics of probabilistic Description Logics, that allows one to provide a probability distribution over worlds/scenarios, and (iii) established heuristics used by humans for concept composition coming from the field of cognitive semantics. Our approach could be useful in many applicative scenarios, ranging from video games to the creation of new movie characters.

Most recent approaches for action recognition from video leverage deep architectures to encode the video clip into a fixed length representation vector that is then used for classification. For this to be successful, the network must be capable of suppressing irrelevant scene background and extract the representation from the most discriminative part of the video. Our contribution builds on the observation that spatio-temporal patterns characterizing actions in videos are highly correlated with objects and their location in the video. We propose Top-down Attention Recurrent VLAD Encoder (TA-VLAD), a deep recurrent neural architecture with built-in spatial attention that performs temporally aggregated VLAD encoding for action recognition from videos. We adopt a top-down approach of attention, by using class specific activation maps obtained from a deep Convolutional Neural Network pre-trained for generic image recognition, to weight appearance features before encoding them into a fixed-length video descriptor with a Gated Recurrent Unit. Our method achieves state-of-the-art recognition accuracy on HMDB51 and UCF101 benchmarks.