This paper argues that today’s mainstream generative AI models do not yet provide the level of culturalization required in the creative and cultural industries (CCIs). Debates on generative AI in cultural domains have focused primarily on productivity, innovation, intellectual property, labor disruption and bias. While these concerns are important, they do not fully capture a more basic issue: cultural production depends on forms of meaning that are historically situated, socially mediated and unevenly distributed across languages, communities and markets. In such contexts, fluency, safety and technical usefulness are not sufficient measures of system quality. The paper discusses how an existing line of thought from localization, cross-cultural design and culturalization research should be carried into the era of generative AI as explicit attention to cultural context, symbolic meaning and interpretive plurality. The CCIs provide a particularly revealing testbed for this claim because they expose the limits of generic AI systems in domains where value depends on tone, memory, symbolism and representation. To ground the argument, the article revisits earlier work on targeted narratives, personalization and adaptive cultural mediation across museums, heritage sites and pilgrimage routes, and uses recent exploratory attempts with widely used generative AI models to show what remains missing. On that basis, the article proposes a framework for culturalized generative AI and outlines an integrated research and governance agenda spanning AI, the humanities and cultural-sector practice.

Scholarly knowledge graphs integrate bibliographic records from heterogeneous sources and therefore require controlled, auditable deduplication. This paper presents OntoDup, an ontology-driven approach that models entity matching as a governed decision process: Matching outcomes are recorded as reified assertions enriched with governance state, evidence, provenance and operational metadata, while a separate operational view is exposed through policy-driven materialization of consumable identity links. We evaluate OntoDup on the DBLP-ACM and DBLP-Scholar benchmarks under two regimes: (i) a pre-blocked setting using the benchmark candidate lists to compare matching methods under a fixed candidate set, and (ii) an end-to-end setting that generates candidates from the graph with DeepBlocker and applies governed triage and materialization. We report operational precision/recall/F1 computed directly on the graph via SPARQL aggregations, characterize governance workload through state distributions, and quantify inference cost for LLM-based matchers via token and latency metadata attached to assertions. For end-to-end evaluation, we anchor operational links against a full positive reference encoded as idealized validations derived from the benchmark labels, enabling analysis of missed positives in terms of governance status and materialization policy. The experiments show that OntoDup enables evaluation at the level of consumable identity links, review workload, and inference cost, revealing operational trade-offs that are not visible from pairwise matching metrics alone.

Scientific collaboration is increasingly needed to address complex research challenges, yet identifying promising partners in the absence of prior co-authorship remains difficult. We present a decision-support pipeline for discovering researchers who have not previously worked together and whose collaboration is unlikely to emerge without deliberate intervention or institutional incentives. The approach leverages document-level semantic representations to estimate proximity between publications, aggregates these similarities at the author level, and surfaces collaboration opportunities that are not evident from the co-authorship graph. To support interpretation by decision makers, a separate LLM module proposes potential joint research directions, which are subsequently annotated with multi-label fields of study. We evaluate the pipeline through an institutional case study, analyzing 7531 publications from 2009 to 2024 using retrospective, temporally shifted windows. While only a small fraction of suggested pairs materialized spontaneously in subsequent periods, the collaborations that do emerge exhibit strong semantic alignment with the computed recommendations (high cosine similarity) and substantial thematic overlap. These results indicate that semantic proximity can act as an early indicator of latent complementarity between researchers without prior ties, supporting intentional institutional mediation and complementing topology-driven approaches that predict links under passive evolution.

Virtual avatars have shown potential as supports in Autism Spectrum Disorder (ASD) interventions, but many existing systems provide largely standardized interactions that do not account for individual variability. This study presents an exploratory evaluation of a virtual puppet system that enables real-time interaction by synchronously transmitting a human model’s movements, facial gestures, and voice to a digital avatar. The system was personalized using each participant’s restricted interests (RIs), identified through a clinical triangulation process involving therapist input, caregiver reports, and observation. After an initial technical validation with 16 neurotypical children, the system was evaluated in a proof-of-concept sample of 11 children with ASD (7 in an experimental group exposed to RI-based personalization and 4 in a control group interacting with a standard interface). Data sources included eye tracking and therapist-completed observational questionnaires. Across sessions, descriptive patterns in gaze fixation and therapist reports suggested that RI-based personalization may help sustain attention to the screen and support engagement with the therapeutic environment relative to non-personalized interaction. Heatmap patterns further indicated that children under the personalized condition visually explored RI-related elements within the scene. This study provides evidence of technical and procedural feasibility and generates hypotheses for future research.

Topic analysis within broad and evolving fields poses great challenges when attempting to be addressed by traditional methods. In response, topic modeling seeks to automate the identification and analysis of underlying themes within several collections of documents in order to synthesize and facilitate the interpretation of their content. The present study applied multiple iterations of the Latent Dirichlet Allocation (LDA) model and the Best-K mechanism in the identification of topics within a volume of 250 news items labeled with “Artificial Intelligence” within a mainstream web portal. Additionally, a large language model (LLM) was implemented to improve interpretation and description for the labeling of found topics. As a result, 9 key topics were identified ranging from the main trends and challenges in AI.

The development of AI, at an above benchmark pace, has become a worldwide worrying issue and a central narrative in the tertiary level learning institutions. Digital transformation, on the other hand, is updating the educational systems, which necessitates effective governance to steer these improvements. This paper provides an integrated overview of these three pillars – AI, digital transformation, and governance – and explores their interplay in reshaping the landscape of higher education. Drawing on key insights from the EUNIS 2024 conference and recent global studies, the paper examines AI’s transformative potential in optimizing learning and administration, the role of digital transformation in enabling scalable and innovative educational services, and the urgent need for governance frameworks to ensure ethical, equitable, and sustainable practices. The analysis highlights current gaps such as the lagging development of regulatory frameworks amid rapid tech progress, and it sheds light on emerging needs, barriers, and innovations (including the use of AI and XR in teaching and learning, strategies for digital readiness, and models for ethical leadership). Of particular interest is the argument provided AI, digital transformation, and governance, which justifies the need for a Global Compact on AI in education that would foster international collaboration and standards. This all-encompassing tracking and mapping of trends, challenges, and opportunities provides higher education institutions a blueprint for strategic decision making concerning technological integration and serves as a guide for policy makers in the reformation of higher learning education.

The numerous pilgrimage routes that cover the European continent entail an untapped potential to offer slow travel experiences with which to promote the development of rural areas that many people travel through, but very few take time to explore. The rurAllure Horizon 2020 project is investigating ways in which the points of cultural and historical interest in the vicinity of the routes could be connected and put on the map for pilgrims, tourists and locals. Most often, this requires investing in new hospitality services, as well as in the restoration and promotion of specific assets. This paper explains how the IT platform created in rurAllure can help in the decision-making that precedes the investments, by revealing the key locations and needs that, properly connected and catered for, would unlock leisure opportunities that cannot be realised otherwise.

This paper presents a novel hybrid search approach to improve information retrieval from the Salesian Polytechnic University website, addressing the challenge of efficiently managing and accessing the growing volume of information. Leveraging virtual assistant technology, the study combines vector similarity and keyword-based techniques to optimize data retrieval. The methodology involves a structured process, including information gathering, architecture design, search execution and analysis of the results. The system architecture consists of three key layers: the intelligent layer, which uses the OpenAI API for query processing; the data layer, which uses the Qdrant database for storage; and the logic layer, responsible for query execution. Two search methods are applied: Vector similarity search, which retrieves data based on contextual relevance, and keyword search with BM25, which sorts documents by keyword relevance. Testing and analysis confirm that the hybrid search method significantly improves the efficiency and accuracy of information retrieval. The results show a significant improvement in the request measures obtained, where the 4 highest percentages were selected to obtain the context from which the answer is derived. The highest similarity values were 5.56, followed by 3.84, the effectiveness of this method in various knowledge areas of the university website. In conclusion, the hybrid search approach presented in this paper offers a promising solution to efficiently retrieve information from the Salesian Polytechnic University website, improve accessibility and ultimately improve user satisfaction.

In today’s complex business environment, inventory management across multiple retailers presents challenges. Traditionally, each company used its own inventory system, leading to communication issues and inefficiencies. Interoperability between systems was hindered by different technologies and protocols. To address these challenges, a multi-company inventory management system was developed, utilizing the gRPC and SSE communication protocols. gRPC offers high-performance communication through remote procedures, while SSE enables asynchronous event transmission. The system aims to enhance coordination and visibility, allowing companies to share real-time information for data-driven decision-making. Using the SCRUM methodology, customer needs were prioritized, delivering features such as user management, product tracking, order generation, and geolocation integration. The architecture comprised Data, Business, and Geolocation layers, ensuring data accuracy and efficiency. Comprehensive testing validated system functionality, including real-time inventory updates, dynamic route generation, and insightful reports. In conclusion, the system successfully improves multi-company inventory management, promoting efficient communication, synchronization, and data-driven decision-making. Rigorous testing ensures reliability and scalability, offering businesses a robust inventory management solution.

In this paper we present the SPRF (Sensorised Pen with Real time Feedback) platform that aims to objectively detect handwriting-related graphomotor disorders in children with Special Educational Needs (SEN), and to treat them through game-guided intervention plans performed with the same sensorised system. We have developed two sensorised devices (wristband and pen) for the SPRF to be able to quantify relevant physical parameters as children perform both diagnostic activities and the subsequent intervention plans. This way, SPRF differentiates from traditional methodologies commonly adopted in non-technologically assisted approaches where well-known graphomotor variables (directly related to the abovementioned physical parameters) are observed by diverse professionals with different criteria, which undoubtedly leads to an undesirable subjectivity. The novelty of SPRF relates to: (i) the ability to suggest tailored intervention plans for each child, based on objective parameter measurement, and (ii) the possibility to instantaneously correct children as they perform that custom-made plan thanks to an intuitive interface (similar to traffic lights) through which therapists can easily identify skills achieved, in progress or failed. The approach has been validated with a real group of SEN children from Ecuador, showing that, in terms of graphomotor skills, children who received on-the-fly feedback during the therapeutic intervention outperformed children not exposed to such real-time corrections.

Accurate and early prediction of breast cancer recurrence is crucial to guide medical decisions and treatment success. Machine learning (ML) has shown promise in this domain. However, its effectiveness critically depends on proper hyperparameter setting, a step that is not always performed systematically in the development of ML models. In this study, we aimed to highlight the impact that this process has on the final performance of ML models through a real-world case study by predicting the five-year recurrence of breast cancer patients. We compared the performance of five ML algorithms (Logistic Regression, Decision Tree, Gradient Boosting, eXtreme Gradient Boost, and Deep Neural Network) before and after optimizing their hyperparameters. Simpler algorithms showed better performance using the default hyperparameters. However, after the optimization process, the more complex algorithms demonstrated superior performance. The AUCs obtained before and after adjustment were 0.7 vs. 0.84 for XGB, 0.64 vs. 0.75 for DNN, 0.7 vs. 0.8 for GB, 0.62 vs. 0.7 for DT, and 0.77 vs. 0.72 for LR. The results underscore the critical importance of hyperparameter selection in the development of ML algorithms for the prediction of cancer recurrence. Neglecting this step can undermine the potential of more powerful algorithms and lead to the choice of suboptimal models.

Cultural heritage is a domain that produces vast amounts of data, but it is also where the meaning of the data is crucially important, particularly to the extent that it refers to people’s opinions, perceptions, and interpretations of their past and their present, or to people’s feelings, preferences, and attitudes [...]

These datasets have been generated as part of rurAllure project (funded by the European Union’s Horizon 2020 Research and Innovation programme under grant agreement no 101004887). Main goal of rurAllure is the promotion of rural museums and heritage sites in the vicinity of European pilgrimage routes: https://rurallure.eu/project/about/

Cultural heritage is one of many fields that has seen a significant digital transformation in the form of digitization and asset annotations for heritage preservation, inheritance, and dissemination. However, a lack of accurate and descriptive metadata in this field has an impact on the usability and discoverability of digital content, affecting cultural heritage platform visitors and resulting in an unsatisfactory user experience as well as limiting processing capabilities to add new functionalities. Over time, cultural heritage institutions were responsible for providing metadata for their collection items with the help of professionals, which is expensive and requires significant effort and time. In this sense, crowdsourcing can play a significant role in digital transformation or massive data processing, which can be useful for leveraging the crowd and enriching the metadata quality of digital cultural content. This paper focuses on a very important challenge faced by cultural heritage crowdsourcing platforms, which is how to attract users and make such activities enjoyable for them in order to achieve higher-quality annotations. One way to address this is to offer personalized interesting items based on each user preference, rather than making the user experience random and demanding. Thus, we present an image annotation recommendation system for users of cultural heritage platforms. The recommendation system design incorporates various technologies intending to help users in selecting the best matching images for annotations based on their interests and characteristics. Different classification methods were implemented to validate the accuracy of our work on Egyptian heritage.

Recurrence is a critical aspect of breast cancer (BC) that is inexorably tied to mortality. Reuse of healthcare data through Machine Learning (ML) algorithms offers great opportunities to improve the stratification of patients at risk of cancer recurrence. We hypothesized that combining features from structured and unstructured sources would provide better prediction results for 5-year cancer recurrence than either source alone. We collected and preprocessed clinical data from a cohort of BC patients, resulting in 823 valid subjects for analysis. We derived three sets of features: structured information, features from free text, and a combination of both. We evaluated the performance of five ML algorithms to predict 5-year cancer recurrence and selected the best-performing to test our hypothesis. The XGB (eXtreme Gradient Boosting) model yielded the best performance among the five evaluated algorithms, with precision = 0.900, recall = 0.907, F1-score = 0.897, and area under the receiver operating characteristic AUROC = 0.807. The best prediction results were achieved with the structured dataset, followed by the unstructured dataset, while the combined dataset achieved the poorest performance. ML algorithms for BC recurrence prediction are valuable tools to improve patient risk stratification, help with post-cancer monitoring, and plan more effective follow-up. Structured data provides the best results when fed to ML algorithms. However, an approach based on natural language processing offers comparable results while potentially requiring less mapping effort.

With so much confusion around use case modeling, software developers may be reluctant to apply it. However, use cases are easy to apply and have great benefits even in research settings with nonprofessional software developers involved. We report on our experience on this within an innovative tourist information system developed in a research setting of an ongoing Horizon 2020 project named Promotion of Rural Museums and Heritage Sites in the Vicinity of European Pilgrimage Routes (rurAllure). The resulting use case model comprises forty use cases supported by class diagrams, use case diagrams, sequence diagrams, as well as by the domain model and dictionary. Three quarters of use cases have been implemented at least to some extent. Writing use cases provoked intense communication, which helped consolidate the system architecture. The use cases helped validate and better understand the GUI form wireframes designed beforehand and get quickly to a consistent implementation of the system that can be experimented with further in different ways. Finally, the forthcoming extensive system and user experience testing will be guided by the use case model. The paper also exposes the domain model and dictionary along with selected use cases and the hierarchy of user and system actors, which may be useful as such or as a stimulus for the development of other innovative tourist information systems.

This paper presents the design and simulation of a driving assistant established by means of a Multisensory Alert System (MAS). MAS allows to warn the driver about imminent risks, the road situation, the environment surrounding the vehicle and other variables of the vehicle. MAS is established as an integral driving assistance system that enables communication between vehicles (V2V) and with access points located on the road (V2I). MAS is composed of several subsystems that trigger auditory, visual and haptic warnings to better engage the driver’s attention according to his or her configuration preferences, resulting in a driving assistant that is much more adaptive to the user’s driving profile. MAS indicates visual warnings through a Human Machine Interface (HMI), auditory warnings through the vehicle’s speakers and haptic warnings through vibration motors located on the steering wheel and accelerator.

Multiple technological aids have been proposed to support the social interactions of people with autism. The literature is extensive, but in many of the works there is a lack of clear processes focused on improving social skills. In this paper, we analyse a selection of 57 technological supports from the last 10 years, whose main characteristic is the quality and clarity of the design and evaluation processes. The objective is to provide the research community with a complete and interdisciplinary mapping of the main technological supports applied to social interaction, identifying the existing relationships between different axes of analysis: the psychological theories that support each approach, the types of technology used and the therapeutic models applied.

The European Union is investing in the areas of digital skills, digital infrastructures, digitisation of businesses, and public services to speed up numerous administrative processes and to facilitate access to citizens from member countries and neighbouring ones as well. This study provides a quantitative assessment of the efficiency gains that can be attained by the ongoing digital transformation in the realm of Erasmus+, the European Commission’s programme for education, training, youth, and sport for the period 2021–2027. This programme manages a sizable budget allocated to education and training opportunities abroad for millions of students, teachers, and other staff of Higher Education Institutions within the EU and beyond. The management of such experiences has significantly grown in complexity over the last decades, entailing notable expenses that the EC aims to reduce through the end-to-end digitalisation of administrative procedures. Our analysis of the savings attained by the so-called Erasmus Without Paper project (EWP) was conducted by taking a close look at the workload, resources, and money invested in Erasmus+ proceedings by four universities from Spain, Italy, and Turkey. The analysis revealed significant savings in terms of paper wastage (a reduction of more than 13.5 million prints every year for the whole Erasmus+ programme) and administrative time, which may translate into lower staff effort and increased productivity, to the point of managing up to 80% more mobilities with the same resources and staff currently available.