Research Challenges in Information Science

Combining data sources from NoSQL and SQL systems leads to data distribution and complexifies user queries: data is distributed among different stores having different data models. This data implementation complexifies the writing of user queries. This work proposes a querying framework of a polystore with the use of unified models as a user vision of the polystore. Unified models hides the variety of data models and data distribution to the user. Our solution uses the Entity-Relationship model of the polystore to infer unified models and to identify intermediate required operations to fulfill querying on real polystore. Using these required transformations, a rewriting framework allows to automatically rewrite the user query (against the unified model) with respect to the real data distribution over the polystore. We apply this framework with one dataset (UniBench benchmark) between a relational, a document-oriented and a graph-oriented databases. We illustrate in this work performance and the low impact of our query rewriting solution when compared to query execution time.

The aim of Open Science is to open up data to enrich knowledge creation processes. At present, Open Science actors face problems when trying to find and exchange data. Research data management platforms need to address the issue of interoperability to enable interdisciplinary research. Some solutions are available for specific communities, but none addresses the problem as a whole. Based on an extension of the theoretical model of interoperability, which enables us to define the criteria for an information exchange, we have quantitatively evaluated information exchange in Open Science. On the basis of this explorative study, we propose an inter-community and inter-disciplinary information exchange network solution enabling decentralised and federated data management as well as a unified search for datasets across all the entities registered in this network: the Open Science Data Network (OSDN). We carried out a proof of concept to assess the feasibility of the solution. We also evaluated this solution by applying it to a completed agronomy research project. This evaluation enabled us to measure a 7% increase in the volume of data, with an 80% reduction in the time needed to find this data. In addition, users have been able to design new intra- and interdisciplinary futures works whit data found.

Retrieving specific information from extensive scientific documents presents a significant challenge. Existing information retrieval methods often focus on entire documents, even when only a small portion of a document is relevant. Also, achieving a balance between precise retrieval and optimal time complexity remains a persistent challenge. To address these issues, we propose TD-CRESTS (Top-Down Chunk Retrieval based on Entity, Section, and Topic Selection), a term-based document text chunk retrieval method. TD-CRESTS utilizes a hierarchical context architecture, indexing documents according to topics, named entities, sections, and individual text chunks. Key terms from each context guide a top-down search strategy across the index context levels, prioritizing the most relevant contexts based on their overlap with the query. Our method achieves chunk-level F1-measure of \(71\%\) and \(77.14\%\) on the SciREX and DrugSemantics benchmark datasets, respectively. It is able to handle diverse domains and languages with a balance between information retrieval efficiency and effectiveness.

Semantic web technologies have been frequently used in smart agriculture systems to address integration and interoperability issues, while enhancing expressiveness through their automatic reasoning capabilities. Despite continuous advancements in the field, there is no existent ontology capable of providing a comprehensive representation of the spatial and temporal knowledge, essential in the context of smart farming. In this paper we propose such a model, pass it through a rigorous validation and verification process, and employ it as a representational layer in an IoT agricultural application. The resulting system utilizes real-world high-quality information to offer various functionalities, including crop health monitoring and disease risk forecasting. This work opens a new perspective regarding the development of smart farming applications, by enabling ontological models to fully exploit the spatial and temporal dimensions of agricultural information.

Events in recent years such as the Russo-Ukrainian war of 2022 and the covid-19 pandemic have once again shown the importance of relying on resilient supply chain networks. The creation and maintenance of such networks is, however, a rather knowledge intensive task, which is still challenging. To tackle this, we introduce a first version of a knowledge graph-based decision support system aiming to help supply chain risk managers to make sourcing decisions. The system was designed by following the design science research methodology, which is supplemented with the Ontology Development 101 [25] for rigor in creation of the knowledge graph schema. Competency questions elicited with domain experts were used to evaluate the proposed system.

In the light of technological advancements, disruptions and regulatory changes, various guidelines and standards emphasize the need for technology resilience. However, they often lack explicit evaluation methods, leaving organizations to determine their own implementation and assessment approaches. This absence of specific guidance amplifies the challenges organizations face in ensuring business continuity when critical systems fail. To address this, a “digital immune system” is proposed – a holistic approach to safeguard digital assets and mitigate IT-related risks. Digital immune system integrates processes, analytics and technologies to strengthen IT architecture, business operations and incorporates assessments to evaluate technology ecosystem resilience. Despite the acknowledged need for technology resilience, existing frameworks fall short in providing practical evaluation methods for the digital immune system. This paper confronts this challenge by focusing on the interconnected components of technologies, data and processes, considering emerging threats and compliance requirements. The goal of the research is to design an assessment framework for the digital immune system and establish a Digital Immune System Maturity Model. The model offers a structured path for organizations to measure and improve their IT risk assessment, resilience and business continuity plans.

Requirements Engineering (RE) stands as the cornerstone in ensuring that a system comprehensively captures and analyzes the needs and expectations of its users and stakeholders. Despite the numerous approaches designed for dealing with functional and quality (non-functional) requirements, approaches for dealing with emotional requirements still lag. Emotional requirements capture how users should feel when using a system, and inadequate consideration of such requirements results in end-users reluctance to use the system. In this paper, we report on our experience in dealing with emotional requirements as part of an H2020 European Project, namely PHArA-ON (Pilots for Healthy and Active Ageing in Europe) for the development of the PHArA-ON ecosystem that aims at improving the well-being and active aging of older adults. Specifically, we present the process we followed for dealing with emotional requirements, and we summarize the findings and lessons learned from this experience.

As the population grows older, the need for special assistance increases, and a modern alternative to mitigate the absence of face-to-face caregivers (which is expensive) is to take advantage of technological devices in so called smart environments, which can be an economical and practical solution. Guaranteeing the software quality of applications in these spaces before providing it to end users is essential, especially in situations involving senior citizens or people with motor disabilities. In order to investigate how the quality evaluation of smart environment applications has been performed, we carried out a tertiary study. From a total of 1,028 studies, 21 were carefully selected for analysis. The results confirmed that classical questionnaires and interviews are the techniques that are still used the most for evaluation, but that simulation appears as a new trend to that end.

Dry Ports are critical infrastructures in the logistics chain, optimizing seaport operations by providing customs services and container storage on land. Some of their main challenges include traffic congestion, environmental impact, high dependence on regulations and paper documentation, and the need to provide reliable information to government authorities and road and rail freight industries. More recently, Digital Twins have emerged as a promising solution to monitor real-time information and optimize Dry Port operations simultaneously. This paper presents the initial proposal of a reference architecture for developing Dry Port’s Digital Twins. It results from a design science research project conducted in cooperation with a major IT company in the logistics sector and a Dry Port operator in Portugal. ArchiMate was the selected language for architecture modeling. Our theoretical contribution is in the form of a high-level reference architecture, confirming the suitability of the ArchiMate language. For practitioners, this work is part of the Portuguese NEXUS agenda: Innovation Pact for Green and Digital Transition for Transport, Logistics, and Mobility, assisting in adopting Blockchain, Optical Recognition, and Artificial Intelligence as crucial technological enablers.

Predictive process monitoring is an evolving research field that studies how to train and use predictive models for operational decision-making. One of the problems studied in this field is that of predicting the sequence of upcoming activities in a case up to its completion, a.k.a. the case suffix. The prediction of case suffixes provides input to estimate short-term workloads and execution times under different resource schedules. Existing methods to address this problem often generate suffixes wherein some activities are repeated many times, whereas this pattern is not observed in the data. Closer examination shows that this shortcoming stems from the approach used to sample the successive activity instances to generate a case suffix. Accordingly, the paper introduces a sampling approach aimed at reducing repetitions of activities in the predicted case suffixes . The approach, namely Daemon Action, strikes a balance between exploration and exploitation when generating the successive activity instances. We enhance a deep learning approach for case suffix predictions using this sampling approach, and experimentally show that the enhanced approach outperforms the unenhanced ones on event logs that exhibits a high frequency of repeated activities with respect to both control-flow and temporal accuracy measures.

Organizations are obliged to ensure compliance with an increasing amount of regulatory requirements stemming from laws, regulations, directives, and policies. As a first step, it is to be determined which of the requirements are relevant in a certain context, depending on factors such as location of the organization and the business processes. For the processes, the identification of relevant requirements can be detailed by an assessment of which parts of each document are relevant for which step of a given process. Nowadays the identification of process-relevant regulatory requirements is mostly done manually by domain and legal experts, posing a tremendous workload due to the extensive number of regulatory documents and their frequent changes. Hence, this work examines how organizations can be assisted in the identification of relevant requirements for their processes based on embedding-based NLP ranking and generative AI. The evaluation highlights strengths and weaknesses of both methods regarding applicability, automation, transparency, and reproducibility. The evaluation results lead to guidelines on which method combinations will maximize benefits for given characteristics such as process usage, impact, and dynamics of an application scenario.

Process discovery aims to discover descriptive process models from event logs. These discovered process models depict the actual execution of a process and serve as a foundational element for conformance checking, performance analyses, and many other applications. While most of the current process discovery algorithms primarily rely on a single event log for model discovery, additional sources of information, such as process documentation and domain experts’ knowledge, remain untapped. This valuable information is often overlooked in traditional process discovery approaches. In this paper, we propose a discovery technique incorporating such knowledge in a novel inductive mining approach. This method takes a set of user-defined or discovered rules as input and utilizes them to discover enhanced process models. Our proposed framework has been implemented and tested using several publicly available real-life event logs. Furthermore, to showcase the framework’s effectiveness in a practical setting, we conducted a case study in collaboration with UWV, the Dutch employee insurance agency.

The discovery of process models from event logs has been a well-understood topic regarding activity-centric processes. For alternative paradigms (e.g., data- or object-centric processes as implemented in many information systems), however, this model discovery still poses several challenges. One of these challenges concerns the discovery of object behavior expressed in terms of object lifecycle processes. In particular, this discovery requires the consideration of different granularity levels (i.e., object states and object attributes). This paper presents an approach for discovering object lifecycle processes. The approach divides the discovery of object lifecycle processes into subproblems by preprocessing event logs to enable the use of well-known discovery algorithms. Overall, object-centric process mining gives insights into data-driven and object-centric processes as implemented in many information systems.

Constant integration of new technologies in our daily lives exposes us to various security threats. While numerous security solutions have been developed to protect us from these threats, they fail due to users’ insufficient comprehension of how to employ them optimally. This challenge often stems from inadequate capture of Usable Security (USec) requirements, leading to these requirements being overlooked or not properly considered in the final solution, resulting in barely usable security solutions. A viable solution is to adeptly capturing USec requirements. Although techniques like User Stories (US) have gained popularity for focusing on users’ needs, they encounter difficulties when dealing with non-functional requirements (NFR), like USec. This occurs due to the lack of well-defined US models explicitly tailored to address these particular requirements. This paper aims to tackle this issue by proposing US4USec, a US model tailored for USec. US4USec has been constructed based on best practices for the consideration and integration of NFR into US models that have been identified via a Systematic Literature Review (SLR). The coverage and completeness of US4USec have been demonstrated by applying it to a set of security US.

Despite recognizing the applicability of Cialdini's principles in social engineering context, studies on their effectiveness needed more tailored and validated tests, primary data collection, and multicultural samples. Cialdini's six persuasion principles include reciprocity, commitment, liking, scarcity, social proof, and authority. We designed and face validated 12 scenarios representing the presence and absence of each principle in a situation where an acquaintance prompts online group members to install an app for testing and improving it. Through an online survey with 314 UK and 328 Arab participants, we collected data on the impact of persuasion principles on risk taking, i.e., to accept installing and trying the app, and trust in the requester, who might be knowingly a social engineer. Results across both cultural frameworks indicate significant impacts, with Social Proof and Authority being the most influential, and Scarcity the least, yet still significant. Interestingly, the principles not only influenced the decision to take the risk but also affected trust in the potential social engineer. This holds true even in less intuitive scenarios, representing Scarcity and Commitment/Consistency principles. This applies to two distinctive cultural frameworks, Arab and British, increasing robustness. The research also investigates the relationship between security attitudes, measured through SA-6 scale, and susceptibility to these principles, in terms of trust and risk taking, revealing surprising results of positive correlations. These findings emphasize the need for cybersecurity strategies that include awareness of psychological manipulation alongside technical knowledge, catering to different cultural contexts.

While cybersecurity is of high relevance for all organizations, special care is needed in the healthcare and social realm when coping with sensitive patient data. This study contributes to this under-investigated yet relevant field by examining how cybersecurity measures have been implemented within healthcare and social organizations. We rely on a combination of clustering analysis, discriminant analysis, and Tukey HSD testing to analyze survey data on 265 organizations in Flanders, Belgium. The resulting five clusters unveil five distinct approaches or organizational profiles and three major differentiators. The data suggests that the extent to which training, regular software updates, and data backup are implemented best describes the underlying cybersecurity profiles. Our findings reveal that a significant majority of surveyed organizations are situated in the lower echelons of the cybersecurity implementation differentiators, while only a minority of organizations demonstrate commendable levels of implementation. By enriching cybersecurity insights within the healthcare and social domain, our findings and their implications could resonate deeply, urging researchers to expand their research to bolster cyber resilience in specific sectors.

Despite the availability of metrics and measurement tools, the lack of formal models and standardization concerning product lifecycle information poses a challenge in assessing how sustainable the supply chain operations of a company are. Digital Product Passports (DPPs) emerge as a promising solution to track and ensure accurate product information is maintained throughout the whole product lifecycle. DPPs are digital representations of the accumulated information of a particular product from its inception to end-of-life. We investigated the topic from the perspective of a large European manufacturer, examining how different phases of the product lifecycle can be supported by static and dynamic data at different levels of granularity. As a result of our research, we propose a reference architecture that supports the development of DPPs with emphasis on product components at batch level granularity. The validation of the proposed architecture shows that the approach provides an opportunity for manufacturers to address sustainability issues in resource-intensive manufacturing supply chains while actively reusing legacy infrastructure.

The presence of Artificial Intelligence and specifically Machine Learning (ML) has increased in all manner of software applications, and it already plays a major role in a variety of systems pertaining to Information Science such as public transport, disease diagnosis support and other medical problems. This increase in use has raised concerns about possible environmental impacts, since ML models require to be trained in datacentres that can impose a high ecological toll. With the aim of uncovering new ways of reducing the energy consumption of ML models, in this study we will explore the energetic impact of class balance for binary classification tasks by comparing a set of logistic regression models (LRMs) trained on a synthetic balanced dataset against another set trained on a synthetic, unbalanced dataset. We focus on the total energy and time required to complete the task, and discover that the order in energy efficiency of the models remained consistent regardless of class balance, but those trained on the unbalanced dataset required between 1.42 and 1.5 times more energy to complete the tasks, despite requiring only around 1 s more of runtime. We finish by analysing the results and proposing using synthetic datasets to estimate the energy cost of different hyperparameter options for LRMs.

[Context] To assess the sustainability performance of companies, diverse environmental, social and governance accounting (ESGA) methods exist, each with their own set of topics and indicators. In earlier research, we have shown that several ESGA methods contain overlapping indicators. [Aim] We aim to develop a semi-automated approach for identifying the overlap between ESGA methods, and then merging the methods into a single combined method that has no redundant indicators. [Method] We have approached this goal as a model management challenge. We have surveyed companies to formulate the problem statement, conducted a literature study on model management operations, created ESGA method models according to our openESEA domain-specific language, and developed algorithms that leverage the power of language models to match and merge the methods. The matching threshold is determined by performing an experiment with 16 experts. Lastly, we validate our algorithms by merging 4 real-life ESGA methods. [Result] The algorithm has proven capable of successfully identifying overlap between ESGA methods. While we would prefer to further reduce the number of false positives, the results already provide valuable insights into the optimisation of sustainability accounting. Moreover, our findings demonstrate how language models can be used for model management.

Current market contexts are pushing leaders to face the need for new business paradigms and to adapt their organisations to an ever-faster changing, dynamic and even unpredictable world. Organisations are requiring new ways of operating and new IT governance approaches that allow them to act quickly while remaining alert to the signals of the environment in which they operate. In this sense, we have focused our work on the design and implementation of an Adaptive Portfolio Management (APM) approach based on complexity and sociotechnical design principles in two Spanish organisations. After some Action-Research cycles, we conducted a situational analysis phase through several focus groups, the results of which are presented in this paper. Our findings show that the sociotechnical approach has brought to light the interrelationship between the governance model and other dimensions such as strategy, operating models, organisational design, people and culture, and technology. One of the key findings of the diagnosis is that, by implementing a governance model devoid of deterministic principles, we can readjust the other dimensions and introduce changes to facilitate the organisation’s ability to adapt to dynamic environments.

Misinformation can harm individuals and societies, with social media and online communities amplifying its reach and impact. One effective strategy to counteract the spread of misinformation online is social corrections, in which people on social media actively challenge others who post or spread it. People hesitate to do so for reasons related to empathy, fear of affecting their relationships, futility, and subjective norms. This research aims to explore the impact of empathy on individuals’ willingness to challenge misinformation. The research also investigates the mediation role of the personal factors of perspective-taking and the need for cognition, along with the perceived impacts on their relationships, on the relationship between empathy and the willingness to challenge. The data was collected from 250 UK-based social networking users and then analyzed using Partial Least Squares Structural Equation Modeling. The results of the analysis supported that perspective-taking (β = 0.064, p = 0.011), the need for cognition (β = 0.022, p = 0.048), and perceived relationship costs (β = 0.035, p = 0.003) all fully mediated the impact of empathy on the willingness to challenge misinformation. The results also show that empathy does not have a direct impact on willingness to challenge misinformation. Individuals with varying levels of empathy converge in their attitudes toward challenging misinformation influenced by a combination of cognitive processes and considerations of their relationships.

Industrial-grade games, like Space Engineers, must adopt swift development and testing processes to conform to rigorous quality standards. Nevertheless, the testing phase of these extensive and complex games heavily relies on manual effort from play-testers, leading to productivity constraints during development cycles. This experience paper reports a Behavior-Driven-Development (BDD) software development process for automated regression test scenarios that allows complement testers’ work during development cycles. To enable BDD test scripts for the Space Engineers game, we have extended the iv4XR framework into a game plugin to connect and execute game actions. Additionally, we have integrated the Cucumber software to describe game test scenarios using natural language. This approach allows testers to create, maintain, and execute a subset of regression test scenarios by relying on a BDD agent that can autonomously verify Space Engineers game features, enabling seamless integration into the development cycle.

In this study, we aim to establish the connection between the emotional trajectory of students during a pedagogical sequence and their performances. The project aims to develop an affective and intelligent tutoring system for detecting students facing difficulties and helping them. We designed this experimentation during the 2022–2023 academic year with students in a French engineering school. We collected and analyzed two primary data sources: student results from the Learning Management System (LMS) and images captured by students’ webcams during their learning activities.

It is known that basic (primary) emotions (like fear or disgust) do not reflect student affective states when facing pedagogical issues (like misunderstanding or proudness). Since such “academic emotions” are not easy to define and detect, we changed the paradigm and used a 2D dimensional model that describes better the wide spectrum of emotion encountered. Moreover, it allows to build a temporal emotion trajectory reflecting the student’s emotional trajectory.

Firstly, we observed a correlation between these trajectories and academic results. Secondly, we found that high-performing student trajectories are significantly different from the others. These preliminary results, support the idea that emotions are pivotal in distinguishing highly performing students from their less successful counterparts. This is the first step to assess students’ profiles and proactively identify those at risk of failure in a human learning context.