Abstract
Metadata and retrieval functions play a vital role in aiding researchers in the discovery and reuse of open data. However, the diversity of metadata elements and retrieval functions poses a challenge to data searchers’ limited attentional resources. This study aims to examine the allocation of attention to metadata elements and retrieval functions and its implications for perceived value and intentions to discover and reuse open data by drawing upon the attentional drift-diffusion model, flow theory, and perceived value literature. An experiment with 48 participants was conducted to explore the proposed relationships. Multiple linear regression analysis was performed to analyze the data. The results suggest that researchers’ attention to high-value functions amplifies the perceived value and motivates data discovery intention. Attention to high-value metadata elements motivates data discovery and reuse intention. In contrast, attention to low-value metadata elements hampers the perceived value and inhibits data discovery and reuse intention. These findings put forward a new lens for exploring the attention mechanisms underlying perceived value, data discovery and reuse intention and highlight the important role of the value of metadata and retrieval functions in attention mechanisms. Additionally, this paper identifies the positive effect of perceived ease of use on users’ intentions to find, evaluate, and access open data. Perceived usefulness positively affects users’ intentions to evaluate open data. However, in contrast to perceived intentions to reuse open data assessed by self-reported measures, perceived value is not a salient motivator of open data reuse intention measured by behavioral indicators. These findings reveal the distinct effects of perceived value on perceived intention and intentional action in data reuse. With these insights, this study develops practical strategies to optimize the design of metadata and retrieval functions in data retrieval systems.
Keywords
Introduction
Open data are basic resources that can play an important role in science and technology innovations (European Commission, 2011) and sustainable development (United Nations Educational Scientific and Cultural Organization, 2015) only when they are reused (Janssen et al., 2012). Open data discovery involves data search, evaluation, and access and is a prerequisite to reusing open data (Gregory et al., 2020b; Liu et al., 2021). Data retrieval systems are critical technical tools that support researchers in discovering and reusing open data. It retrieves data according to query and returns a list of relevant datasets sorted by certain rules, resource links, and data descriptions such as metadata and data documents (Kacprzak et al., 2017). Metadata and retrieval functions are key to data retrieval systems. Specifically, metadata describe the key characteristics of data (e.g. title, abstract, source, and citation frequency). Additionally, data retrieval systems provide users with functions such as keyword retrieval, sorting, filtering, and search result analysis (Khalsa et al., 2018; Neves et al., 2018). These can support researchers in searching (Wu et al., 2019), evaluating (Krämer et al., 2021), accessing (Koesten and Singh, 2017), and understanding (Pasquetto et al., 2019) data and then reusing it. However, the diversity of metadata and retrieval functions may lead to the attention-allocation problem. Since information consumes recipients’ attention, a wealth of information may result in a poverty of attention, resulting in a need to efficiently allocate attention (Simon, 1971).
Problem statement
Information search has been regarded as an interactive process in which human attention engages (Belkin and Croft, 1992; Marchionini, 2004). Various metadata elements and retrieval function pose a challenge to data searchers’ limited attentional resources. In the field of cognitive psychology, the attentional drift-diffusion model points out that attention affects individual value perceptions (Orquin and Mueller Loose, 2013). However, to the best of our knowledge, no previous study has investigated data searchers’ attention allocation among metadata and retrieval functions and its impact on their perceived value. Perceived value studies note that value is formed in interaction and reflects users’ evaluations of products or services (Sánchez-Fernández and Iniesta-Bonillo, 2007). Perceived value is found to affect open data discovery and reuse intentions (Talukder et al., 2019). Moreover, flow theory suggests that focused attention affects intention (Liu and Zhou, 2017). However, little is known about the relationships between attention allocation and researchers’ intentions to discover and reuse open data. It is necessary to examine the relationships between researchers’ allocation of attention to diverse metadata and retrieval functions, perceived value, and data discovery and reuse intention. Therefore, this study uncovers the implications of the diversity of metadata and retrieval functions for researchers’ attention allocations, perceived value, data discovery and reuse intention by drawing on perceived value studies, the attentional drift-diffusion model, and flow theory. This study aims to answer the following research questions:
During the interaction with the data retrieval system,
RQ1. What is the relationship between the allocation of attention to metadata and retrieval functions and perceived value?
RQ2. What is the relationship between the allocation of attention to metadata and retrieval functions and open data discovery and reuse intention?
RQ3. What is the relationship between perceived value and open data discovery and reuse intention?
To answer these questions, 48 participants were recruited to participate in a user experiment. We asked the participants to complete four search tasks in the data retrieval system. We then collected their search logs, screen recordings, questionnaires, and eye movement data and used stepwise regression analysis to analyze the data.
The main findings and their theoretical contributions can be summarized as follows: First, it provides new insights into researchers’ attention allocations among diverse metadata elements and retrieval functions and its implications for the perceived value of data retrieval systems. The findings suggest that researchers’ attention to high-value metadata elements and functions amplifies the value they perceive, while attention to low-value metadata elements hampers the perceived value. By highlighting the importance of attention allocation in perceived value during interactive data retrieval, these findings advance the current understanding of perceived value and the role of attention in interactive retrieval. Second, this study sheds light on the motivating and inhibiting effects of attention allocation on data discovery and reuse intention. Specifically, researchers’ attention to high-value functions motivates their data discovery intentions. Attention to high-value metadata elements motivates data discovery and reuse intention, while attention to low-value metadata elements is an inhibitor. It advances the understanding of the cognitive processes underlying data discovery and reuse intention from a novel perspective of attention mechanisms. It also expands the knowledge of the effect of attention on behavioral intention by highlighting the importance of the actual value of gazed objects. Third, this study extends the existing knowledge of the motivating mechanisms underlying perceived intention and intentional action in open data reuse by uncovering the distinct effects of perceived value on perceived intention and intentional action. Previous studies have used self-reported measures to assess users’ intentions to reuse open data and identified the positive effects of perceived value on perceived data reuse intention. This study uses behavioral indicators to investigate open data reuse intention. It identifies that perceived value is not a salient motivator of data reuse intention, as revealed by behavior. Drawing upon intentional action theory, it uncovers the differences between perceived intention and intentional action in open data reuse and the distinct motivations underlying them.
Literature review
Open data discovery and reuse involve open data needs, search, evaluation, access, sensemaking, and reuse, among which open data needs, search, evaluation, and access are the keys to open data discovery, and open data sensemaking and reuse are the keys to open data reuse (Gregory et al., 2019; Kim and Yoon, 2017; Koesten et al., 2017). Previous studies suggested that data retrieval system metadata and retrieval functions (Koesten and Singh, 2017; Wu et al., 2019) are important in open data discovery and reuse. For instance, keyword-based retrieval and browsing functions (Kern and Mathiak, 2015) and metadata have been found to be critical factors for researchers to successfully search data (Löffler et al., 2021). Researchers can save data through data links and download functions (Bugaje and Chowdhury, 2017).
Cognitive processes involved in the use of data retrieval systems for data discovery and reuse are complex (Koesten et al., 2017; Pasquetto et al., 2019). For instance, based on metadata and data documents, researchers evaluate data relevance (Kern and Mathiak, 2015), fitness of data for use (Bishop et al., 2019), availability and quality (Koesten et al., 2017; Krämer et al., 2021), and credibility (Gregory, 2020) and obtain knowledge about data and data creation process to make sense of it (Pasquetto et al., 2019). Several data retrieval systems support interactive data exploration (Devarakonda et al., 2021), which can support researchers in understanding and reusing data (Gregory et al., 2019; Koesten et al., 2021). During these processes, it’s challenging for researchers to allocate limited attention resources to various metadata elements and retrieval functions effectively. However, few studies have revealed the relationships between metadata and retrieval functions, system evaluation, and open data discovery and reuse from a cognitive perspective, especially from an attention perspective.
Additionally, self-reported measures such as questionnaires (Talukder et al., 2019; Weerakkody et al., 2017) and interviews (Kaasenbrood et al., 2015) are used by researchers to investigate open data reuse intention. Rzepka and Berger (2018) noted that there are differences between self-reported perception and pattern or tendency revealed by behavior in the process of user interaction with the system. Thus, this study develops behavioral indicators to investigate open data reuse intention.
Theoretical background
Perceived value
Value is an overall assessment of the utility of a product or service given by consumers according to the perceptions of receiving and giving (Zeithaml, 1988). Perceived value is an interactive relativistic preference generated during a subject’s interaction with a product or service (Holbrook, 1999), which is individualistic and relativistic (Gallarza et al., 2011; Yu et al., 2017). Experience is the essence and basis of perceived value (Hsiao et al., 2016; Turel et al., 2007). Together, value is the subjective evaluations of products or services by subjects based on their interaction with the products or services. In this study, perceived value is regarded as users’ subjective evaluations through their interaction with the data retrieval system.
Perceived value is considered a multidimensional construct that includes functional value, emotional value, social value, epistemic value, and conditional value (Sheth et al., 1991). Functional value and emotional value are the most widely studied. Specifically, functional value comes from a product’s attributes or characteristics (Hou et al., 2020). Emotional value depends on the feelings generated by products or services (Sánchez et al., 2006). In information retrieval systems, usefulness is the perception that the system can support users in completing tasks and improving search efficiency. Ease of use is the perception that systems can help users reduce mental effort to support interaction with the system (Tella et al., 2017). These two factors are closely related to functional value. Therefore, this study explores data searchers’ evaluations of functional value from perceived usefulness and perceived ease of use. Research on emotional value mainly explores users’ positive emotions during interactions with products or services (Sweeney and Soutar, 2001; Zhou, 2008). However, there are few empirical studies on user emotion in the process of data retrieval, and most of them focus on negative emotions (Megler and Maier, 2012; Wiegand and García, 2007). This study investigates users’ positive emotions (i.e. perceived enjoyment) during their interactions with a data retrieval system to understand researchers’ evaluations of emotional value.
Perceived value has been found to affect behavioral intention (Osakwe et al., 2021; Wang et al., 2013). Perceived usefulness and perceived ease of use have been found to affect open data discovery (Kim and Yoon, 2017; Zuiderwijk et al., 2015) and reuse (Jurisch et al., 2015; Weerakkody et al., 2017) intentions. Perceived enjoyment has been found to affect open data reuse intention (Khurshid et al., 2022). However, few studies have investigated the role of perceived value in interactive data retrieval. Driven by this gap, this study explores users’ perceived value and its impacts on open data discovery and reuse intention in the context of researchers’ interaction with data retrieval systems.
Attentional drift-diffusion model and the flow theory
Attention is a critical cognitive process (Léger et al., 2014; Posner, 2012), indicating the selection and further processing of external stimuli (Eysenck and Keane, 2020; Hilberink-Schulpen et al., 2016), which is regarded as an important and limited resource for humans (Cheung et al., 2017). In the process of attention resource allocation, individuals will have an attention preference for some cues, which affects users’ perceived value and choice (Smith and Krajbich, 2019). The attentional drift-diffusion model reveals that attention increases evidence accumulation in decision-making and then increases the decision maker’s perceived value for the option (Mormann and Russo, 2021). The additive model and multiplicative model uncover the distinct mechanisms underlying this effect. In the additive model, attention will enhance the evidence accumulated for the gazed-at option, the magnitude of which is irrelevant to the value of the option itself (Cavanagh et al., 2014). There is an interactive effect of attention and the value of the option itself on evidence accumulation in the multiplication model. The amplified effect of gaze on evidence accumulation is affected by the value of the option itself (Krajbich et al., 2010). Attention positively affects perceived usefulness (Cho et al., 2020), perceived ease of use (Tzafilkou and Protogeros, 2017), and perceived enjoyment (Maughan et al., 2007).
Regarding the effect of attention on behavioral intention, flow theory holds that in the flow state, individuals can focus their attention on the current activity and exclude and filter other irrelevant perceived content (Csikszentmihalyi, 1988). Users enjoy the flow state, which has a positive effect on subsequent behavioral intentions (Kazancoglu and Demir, 2021; Kim and Hall, 2019). Focused attention and concentration are important dimensions and characteristics of flow (Pelet et al., 2017). In addition, focused immersion reflects the extent of involvement of individual attention resources in specific activities (Agarwal and Karahanna, 2000), which has been found to be a significant predictor of user behavioral intentions (Venter and Swart, 2018). Attention is considered an important factor affecting information search (Belkin and Croft, 1992; Marchionini, 2004). However, to the best of our knowledge, no study has analyzed the relationships among users’ attention allocations and perceived value and open data discovery and reuse behavior in a data retrieval system. Therefore, this study investigates user attention allocations to metadata and retrieval functions in data retrieval systems and their relationships with perceived value and open data discovery and reuse intention.
Hypothesis development
To address our research problems, we formulate our research hypotheses accordingly. Figure 1 shows the theoretical model.

Theoretical model.
Attention is positively related to perceived value (Krajbich et al., 2010; Mormann and Russo, 2021). Specifically, when users focus on information and communication technology tools, they may have higher immersion when they use these tools and perceive the tools to have higher usefulness (Gloria and Achyar, 2021; Zhou and Lu, 2011). In addition, individuals focus their attention on tasks that allow for acquisition of more cognitive resources, leading to a reduction in cognitive load associated with technology use and thus improving the perceived ease of use of information technology (Agarwal and Karahanna, 2000; Saadé and Bahli, 2005). Moreover, concentration reflects the extent to which an individual’s attention is immersed in activities (Pelet et al., 2017), and it is found to be positively related to perceived enjoyment (Liu and Li, 2011; Wang and Lee, 2020). In data retrieval systems, metadata and retrieval functions are key to supporting data searchers’ search tasks (Koesten et al., 2017). When researchers focus on metadata and the functions of data retrieval systems, they are more likely to realize the support provided by the data retrieval systems for the data search task. The more cognitive resources that are allocated to the task, the higher the perceived usefulness and ease of use. In addition, when researchers allocate more attention to metadata and retrieval functions, they experience less interference from task-irrelevant or low-related stimuli, arousing emotional reactions. Therefore, we propose the following hypothesis:
Perceived value is found to positively affect behavioral intention (Turel et al., 2007). Several empirical studies have found that perceived usefulness and perceived ease of use are positively related to open data reuse intention (Husin et al., 2019; Weerakkody et al., 2017). In addition, perceived enjoyment is positively related to open data use intention (Khurshid et al., 2022). Therefore, we propose the following hypothesis:
Perceived usefulness and perceived ease of use were found to be critical predictors of the use of open data technology (Saxena and Janssen, 2017; Zuiderwijk et al., 2015). Moreover, perceived enjoyment positively affects the intention to use open-access data repositories (Wen et al., 2021). Therefore, perceived value may positively affect the intention to use data retrieval systems. Data searchers use data retrieval systems to discover data. It implies that perceived value promotes the intention to use data retrieval systems, which in turn stimulates the intention to discover data. Therefore, we propose the following hypothesis:
Attention is found to positively affect behavioral intention (Fei et al., 2021). Concentration positively affects the intention to use information systems (Liu et al., 2009). Furthermore, as an important dimension of cognitive absorption, attention was found to positively affect the intention to use innovative technology (Chandra et al., 2009). Therefore, attention may positively affect the intention to use data retrieval systems. Data retrieval systems support users in discovering (Koesten and Singh, 2017) and reusing data (Bishop et al., 2019). Therefore, we propose the following hypothesis:
Method
To test research hypotheses, this study used a controlled experiment that examined an interactive data search. In this section, we present the experimental participants, platform, apparatus, tasks, measures, and procedure.
Participants
As early-stage researchers, graduate students have been found to be important data searchers (Gregory et al., 2020a). Therefore, we recruited 48 doctoral students to participate in the experiment. Among the 48 participants, 45.8% were female and 54.2% were male. A total of 12.5% of the participants were 20–24 years old, 75% were 25–30 years old, and 12.5% were over 30 years old. The participants included science (31.3%), engineering (29.2%), social science (25%), medicine (6.2%), and humanities students (8.3%). All participants had research experience; half of the participants had 4–6 years of research experience, 39.6% had 1–3 years of research experience, and 10.4% had 7–9 years of experience. The experiment participants were asked to complete four tasks and fill out questionnaires during the experiment.
Platform and apparatus
Platform
The Data Citation Index (DCI) of the Web of Science platform was selected as the experimental system. As a data citation database, it can provide professional data retrieval services and a real and stable system environment.
Apparatus
In this experiment, a laptop equipped with a Tobii Pro X3-120 eye tracker was used to capture the screen and record users’ interaction with DCI. The device is at a 120 Hz sampling rate, with 5-point calibration, mounted at the bottom of the screen. Before the experiments started, the participants were asked to adjust themselves to a comfortable sitting position and stay 60–80 cm away from the device during the experiment.
Task
COVID-19 is a common and pressing issue facing human beings. To address this challenge, researchers have openly published research data (Shuja et al., 2021). At the same time, there is a need to discover and reuse open data for COVID-19 research (Ulahannan et al., 2020). It provides a suitable context for exploring researchers’ data discovery and reuse. Therefore, this study designs four data search tasks related to COVID-19 research. The topics of the four tasks are the major topics of COVID-19 studies: clinical characteristics of COVID-19, population mobility, genome sequencing, and psychological and economic issues.
Measures
Attention
This study used eye-tracking data to measure the participants’ attention. Fixation can be used to measure users’ attention (Rayner, 1998), and fixation duration is one of the most commonly used measures of attention (Lee and Ahn, 2012; Pieters and Wedel, 2004). This study used a Tobii Pro X3-120 Eye Tracker to record users’ fixation duration on the following areas of interest (AOIs) to measure the participants’ attention to metadata and retrieval functions: eight metadata AOIs (i.e. title, source, DOI, type, abstract, subject category, publication date, and citation frequency and usage frequency) and five retrieval function AOIs (i.e. adding to marked list and exporting records, filtering and sorting search results, search results analysis, viewing data, and query information) (Figure 2).

An example of 13 AOIs (SERPs and description pages of search results).
Perceived value
Questionnaire data were used to measure perceived value. Perceived usefulness was measured using a three-item scale adapted from Venkatesh et al. (2003) and Chiu and Wang (2008). Perceived ease of use was measured using a four-item scale adapted from Davis (1989). Perceived enjoyment was measured using a three-item scale adapted from Khatri et al. (2018) and Davis et al. (1992). All measures used 7-point Likert scales from 1 (strongly disagree) to 7 (strongly agree).
Open data discovery and reuse intention
The number of documents saved (Belkin et al., 2001; Liu et al., 2012; Zhang et al., 2015) indicates searchers’ usefulness judgments (Liu, 2021). In a data search, the number of documents (i.e. open data) saved by users indicates their usefulness judgments regarding open data. In addition, saving data is the first step in data reuse (Bugaje and Chowdhury, 2017). This study collected log files and screen recording data to obtain this indicator to measure open data reuse intention. In addition, we developed three items in the questionnaire, including open data finding intention, open data evaluating intention, and open data accessing intention, to measure open data discovery intention.
Procedure
Prior to the experiment, we introduced our experiment and required all participants to give informed written consent. Each participant was asked to complete a questionnaire with basic information. Each participant was asked to complete four simulated tasks using DCI. Then, the eye tracker was calibrated using a 5-point standard calibration procedure for each subject. For each task, the participants read the task description and completed a presearch questionnaire. Then, the participants performed the task, each lasting up to 10 minutes. During the task, the participants were able to save data in any way preferred. After completing the task, the participants were asked to complete a postsearch questionnaire. The participants completed four simulated tasks and completed an overall assessment questionnaire. The Tobii Pro X3-120 eye tracker recorded each participant and the entire process. To eliminate the learning effect and fatigue effect, the task order was rotated based on the basic 4 × 4 Latin square (Li and Liu, 2019). The experimental procedure is displayed in Figure 3.

Experimental procedure.
Data analysis
Stepwise regression is a set of iterative search and model comparison processes during which the independent variables that have the strongest associations with the dependent variable can be identified from a large number of alternative variables (Henderson and Denison, 1989; Ma et al., 2016; Vlachopoulou et al., 2013). To support open data discovery and reuse, data retrieval systems make efforts to develop diverse metadata and retrieval functions. Stepwise regression analysis is suitable for identifying the most influential metadata elements and retrieval functions on researchers’ attention allocation and intentions to discover and reuse open data. Perceived value is a multidimensional construct. Stepwise regression analysis is also an appropriate approach to reveal the most influential perceived value on researchers’ intentions to discover and reuse open data. Therefore, to test these proposed relationships, this study performed stepwise regression analysis in SPSS version 26.
Result
The Cronbach’s alpha coefficients for perceived usefulness, perceived ease of use, and perceived enjoyment were 0.959, 0.906, and 0.986, respectively, which exceeded the 0.7 benchmark (Hinkin, 1998), indicating internal consistency. Descriptive statistics of all variables are shown in Table 1.
Descriptive statistics of variables.
As shown in Table 1, researchers allocate most attention resources to data title (M = 38.24, SD = 21.66), followed by data abstract (M = 30.31, SD = 29.50), filtering and sorting search results (M = 29.14, SD = 27.77), adding to marked list and exporting records (M = 19.21, SD = 13.64), data source (M = 16.51, SD = 9.21), query information (M = 13.26, SD = 14.83), DOI (M = 3.84, SD = 1.78), citation frequency and usage frequency (M = 2.70, SD = 2.39), data type (M = 2.39, SD = 1.51), subject category (M = 1.43, SD = 1.33), publication date (M = 1.22, SD = 1.04), and search results analysis (M = 0.68 SD = 0.81). The least attention resources are allocated to the function of viewing data (M = 0.31, SD = 0.65).
The effect of attention to metadata and retrieval functions on perceived value
Stepwise regression analysis, in which the dependent variables were perceived value and the independent variables were participants’ attention to metadata and retrieval functions, was performed (Table 2). The variance inflating factors of all variables in all models (min = 1.000, max = 1.113) were less than 2, indicating the absence of multicollinearity. The Durbin–Watson statistic for all models (min = 1.955, max = 1.995) was between 1.5 and 2.5, indicating the independence of observations. According to the model F test, all models were statistically significant.
Multiple linear regression: perceived value (independent variable: attention measured by fixation duration).
B: coefficient of nonstandard regression; SE: B standard errors; β: coefficient of standard regression.
According to Table 2, attention to the subject category of data has a negative effect on perceived usefulness. However, attention to the query information and the search results analysis function in SERPs are positively related to perceived usefulness. Thus, H1(a) is partially supported. Attention to the data abstract negatively impacts perceived ease of use. Thus, H1(b) is rejected. Attention to the query information has a positive effect on perceived enjoyment. Thus, H1(c) is supported.
The effects of attention to metadata and retrieval function on the intentions to discover and reuse open data
Stepwise regression analysis, in which the dependent variables were the intention to reuse open data and the intention to discover open data and the independent variables were the participants’ attention to metadata and retrieval functions, was performed (Table 3). Regarding the stepwise models in which dependent variables were the intention to reuse open data, the variance inflating factors of all variables in all stepwise models (min = 1.085, max = 1.256) were <2, indicating the absence of multicollinearity. The Durbin-Watson statistic was 2.179, indicating the independence of observations. According to the F test, all models were significant.
Multiple linear regression: the intention to reuse open data and the intention to discover open data (independent variable: Attention measured by fixation duration).
B: coefficient of nonstandard regression; SE: B standard errors; β: coefficient of standard regression.
Regarding the stepwise models in which the dependent variables were the intention to discover open data, the variance inflating factors of all variables in all stepwise models (min = 1.035, max = 1.192) were <10, indicating the absence of multicollinearity. The Durbin–Watson statistic for all models (min = 1.759, max = 1.937) was between 1.5 and 2.5, indicating the independence of observations. According to the model F test, all models were statistically significant.
According to Table 3, the attention to the data title and the data citation frequency and usage frequency positively affects the intention to reuse open data. However, attention to the DOI is negatively related to the intention to reuse open data. Therefore, H4 is partially supported. Attention to the data type and the query information function positively influence the intention to find open data. However, attention to the subject category negatively impacts the intention to find open data. The attention to the data type and search results analysis function positively impact the intention to access open data. However, attention to the data abstract has a negative effect on the intention to access open data. Therefore, H5 is partially supported.
The effect of perceived value on the intention to reuse data and the intention to discover data
Stepwise regression analysis, in which the dependent variables were the intention to reuse open data and the intention to discover open data and the independent variable was perceived value, was performed (Table 4). Regarding the stepwise models in which the dependent variables were the intention to reuse open data, as the results are not significant, the results of the variance inflating factors of all variables and the Durbin-Watson statistic are not reported here. Regarding the stepwise models in which the dependent variables were the intention to discover open data, the variance inflating factors of all variables in all stepwise models (min = 1.000, max = 2.357) were less than 10, indicating the absence of multicollinearity. The Durbin-Watson statistic for all models (min = 1.889, max = 2.433) was between 1.5 and 2.5, indicating the independence of observations. According to the model F test, all models were statistically significant.
Multiple linear regression: The intention to discover data.
B: coefficient of nonstandard regression; SE: B standard errors; β: coefficient of standard regression.
According to the Table 4, perceived usefulness, perceived ease of use, and perceived enjoyment have no significant effect on the intention to reuse data, rejecting H2(a), H2(b), and H2(c). Perceived usefulness is positively related to the intention to find open data, supporting H3(a). Perceived ease of use is positively related to the intention to find, evaluate, and access open data, supporting H3(b). Perceived enjoyment has no significant effect on the intention to discover data, rejecting H3(c).
Discussion
This paper aims to test the relationships among attention to metadata and retrieval functions, perceived value, data discovery intention, and data reuse intention. We proposed our research model and used stepwise regression analysis to test these hypotheses.
Users’ attention to specific retrieval functions (i.e. query information and search results analysis) is found to be positively associated with perceived value. This finding supports previous findings on the positive effect of attention on perceived usefulness (Gloria and Achyar, 2021). Moreover, when users have longer fixation durations for the query information, they may have high perceived enjoyment. This finding aligns with the research finding that attention focus positively impacts perceived enjoyment (Liu and Li, 2011; Wang and Lee, 2020). However, attention to subject category negative affects perceived usefulness, and attention to data abstract negatively affects perceived ease of use. This result may be explained by the multiplicative model of the attentional drift-diffusion model. According to this model, the effect of attention on the perceived value of the looked-at option is influenced by the actual value of the looked-at option itself (Krajbich et al., 2010). The quality of metadata in current open data practice was found to be relatively low (Marc et al., 2016). Users have been found to put much cognitive effort into understanding the metadata of open data (Bugaje and Chowdhury, 2017). The subject category may be too broad and vague to obtain useful information for data evaluation, while data abstracts may be too complex to be easily understood. Our findings suggest that the actual value of these two metadata elements (i.e. subject category and data abstract) is relatively low, which eliminates the promoting effect of attention on perceived value, resulting in more effort and less perceived benefit for users.
Regarding the effects of data searchers’ attention to metadata and retrieval functions on their intentions to discover and reuse data, attention to specific elements of metadata (i.e. data title and citation and usage frequency) significantly and positively affects the intention to reuse data. Attention to specific elements of metadata (i.e. data type) and specific retrieval functions (i.e. query information and search results analysis) positively affects the intention to find and access data. These results support and expand previous findings on the positive effect of attention on system adoption (Liu et al., 2009; Webster et al., 1993). However, attention to specific elements of metadata (i.e. subject category) is negatively correlated with the intention to find data. Attention to data abstracts is negatively correlated with the intention to access data. Attention to the DOI element negatively affects open data reuse intention. This finding is inconsistent with previous findings on the positive effect of attention on behavioral intention (Chandra et al., 2009). These findings may also be explained by the multiplicative model of the attentional drift-diffusion model. Smith and Krajbich (2019) noted that the interaction between the value of the option itself and attention positively impacts evidence accumulation in the decision-making process. Specifically, the decision maker’s gaze on the higher value option has a greater impact on choice (Smith and Krajbich, 2019). Our findings suggest that the actual value of these metadata elements (i.e. subject category, data abstract, and DOI) is relatively low, which eliminates the promoting effect of attention on choice related to data discovery and reuse, leading to a reduction in data searchers’ intentions to discover and reuse data. This implies the interaction effect of the actual value of metadata and attention to metadata on open data discovery and reuse intention.
Perceived usefulness and perceived ease of use are found to be important factors positively influencing the intention to discover data. This finding is consistent with the research results of Saxena and Janssen (2017). However, this study did not find any significant relationship between perceived enjoyment and researchers’ intentions to discover data. This result is inconsistent with previous findings that perceived enjoyment is an important driver of information seeking (Savolainen, 2012). A possible explanation may be that the difference in research contexts leads to the elimination of perceived enjoyment’s motivating effect. In addition, surprisingly, the effect of perceived value on data reuse intention is statistically insignificant. This finding is inconsistent with those of other studies (Khurshid et al., 2022; Talukder et al., 2019). This result may be explained by action theories. Intentional action is action driven by intention, which can reflect an actor’s intention (Burks, 2001). Intentional action is mainly measured by the correlation between intention and subsequent behavior (Burks, 2001). This study used the number of documents saved as a behavioral indicator to measure open data reuse intention. Saving open data documents is a prerequisite to reusing open data (Bugaje and Chowdhury, 2017). This indicates that open data reuse intention is closely correlated with data saving behavior. Therefore, saving open data is an intentional action driven by open data reuse intention and directed toward reuse intention. Perceived value has been found to be an antecedent of intention (Liang et al., 2021). However, intentional action has been found to be affected by self-efficacy, motivation, will power, social pressures, outside resources, social norms, emotional state, and the concept of self as the active agent (Burks, 2001). Our findings suggest that perceived value is not a salient antecedent of intentional behavior in the context of open data reuse.
Theoretical implication
This study reveals the effects of researchers’ attention allocations among metadata elements and retrieval functions on their perceived value of data retrieval systems and the role of the actual value of metadata and retrieval function in these effects. It offers novel insights into the implication of the diversity and quality of metadata and retrieval functions for researchers’ attention allocations and perceived value. Specifically, during interaction with data retrieval systems, researchers allocate their attention across various metadata elements and retrieval functions. This study uncovers the positive effects of data searchers’ attention to specific retrieval functions on their perceptions of value and the negative effects of attention to specific metadata elements on perceived value. These findings support the multiplicative model of the attentional drift-diffusion model, which reveals the interaction effect of attention and the actual value of the looked-at option itself on perceived value (Krajbich et al., 2010). Researchers’ attention to high-value functions amplifies the perceived value of data retrieval systems. In contrast, attention to low-value metadata elements hampers the perceived value. These findings suggest that attention allocation among diverse metadata and retrieval functions significantly affects perceived value. The actual value of metadata and retrieval functions determines the direction of attention effects. Our findings highlight the importance of attention allocation in perceived value during interactive data retrieval, expanding the current understanding of perceived value and the role of attention in information retrieval.
This study uncovers the attention mechanisms in data discovery and reuse intention by analyzing the effect of attention allocation on open data discovery and reuse intention. The role of the actual value of metadata and retrieval functions in these mechanisms is identified. It puts forward a new lens for exploring cognitive processes underlying data discovery and reuse intention. Specifically, attention to the elements of title and citation frequency and usage frequency positively affects data reuse intention, while attention to the DOI element negatively affects data reuse intention. Attention to the data type element and specific functions positively affects data discovery intention, while attention to subject category and abstract elements negatively affects data discovery intention. According to flow theory, focused attention promotes behavioral intention. Our findings suggest that attention can serve as both a motivator and an inhibitor for data discovery and reuse intention. The actual value of metadata and retrieval functions is a determinant of the direction of attention effects. It contributes to existing knowledge of the impact of attention on behavioral intention by highlighting the importance of the quality of gazed objects.
This study reveals the distinct effects of perceived value on researchers’ intentions to discover and reuse open data. The findings identify the positive effect of perceived ease of use on researchers’ intention to find, evaluate, and access open data and the positive effect of perceived usefulness on researchers’ intentions to evaluate open data. This suggests a role for perceived ease of use in promoting open data discovery intention. In addition, we use behavioral indicators to reveal open data reuse intention, which provides a new lens for understanding researchers’ intentions to reuse open data and intentional action. Inconsistent with previous findings of the positive effects of perceived value on open data reuse intention, our findings suggest that perceived value is not a salient motivator of intentional action. It sheds light on the distinct motivating mechanisms underlying perceived intention and intentional action in open data reuse.
Practical implications
This study identifies several low-value metadata elements (i.e. subject category, data abstract, and DOI) and proposes optimizing strategies. Specifically, attention to data abstract negatively affects perceived ease of use. Thus, we suggest that data retrieval systems should make data abstract easier to understand. There is evidence that providing adequate information about the data creation process and data content is the key for users to understand data (Gebru et al., 2021). Structured abstract formats, including elements such as data creation motivation, background, data collection and processing, data content, and recommended use, can be adopted to improve data abstract quality and comprehensibility. Additionally, users’ attention to subject category negatively affects perceived usefulness, which indicates that it is important to improve the usefulness of subject categories in supporting data evaluation. Controlled vocabularies can be used to effectively represent the topic of metadata (Marc et al., 2016), and topic-specific controlled vocabularies can help users search for domain-specific content (Chipangila et al., 2021). This study suggests that a data subject-controlled vocabulary should be designed for subject-specific data searchers so that they can accurately judge the position of data in the subject knowledge system or data resource system. Furthermore, attention to DOI negatively affects data reuse intention. This suggests that the design of DOIs should be optimized to improve users’ willingness to access data. The DOI label should be made easier to understand by changing it to a more intuitive download link label to support data access.
Previous studies have mainly focused on improving the quality of metadata and retrieval functions based on users’ needs (Wu et al., 2019). This study provides a new vision for designing and optimizing metadata and retrieval functions by highlighting the role of users’ attention in the diversity of metadata and functions. The results indicate that data searchers’ attention to high-value metadata elements and retrieval functions (i.e. data type, data title, citation frequency and usage frequency, query information, and search results analysis) is positively related to perceived value, data discovery intention, and data reuse intention. Individuals pay more attention to visually salient options (Kahn, 2017). Data retrieval system designers can improve the visual salience of high-value retrieval function icons and metadata elements by increasing their brightness and color to direct and retain users’ attention.
Conclusion
This paper comprehensively investigates user attention to metadata and retrieval functions and its impact on perceived value and open data discovery and reuse intention during interaction with data retrieval systems. To test the research hypotheses, a controlled experiment was designed to collect data, and stepwise regression analysis was used to analyze data. The findings show that users’ attention to metadata and retrieval functions significantly affects perceived value and open data discovery and reuse intention. Perceived value affects open data discovery intention.
Despite these contributions, the present study has several limitations. First, in this study, the number of documents saved is used as a behavioral indicator to measure open data reuse intention. However, other behavior indicators, such as the number of bookmarks and average bookmark rank, may be suitable for measuring open data reuse intention. Future studies can collect and analyze various behavioral indicators to fully investigate open data reuse intention. Second, although there are various types of data retrieval systems, such as data search engines, data repositories, and open government data platforms, this study chose DCI as the experimental system. Further investigation can examine our research findings in other types of data retrieval systems.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China [No. 72074171].
