Financial Natural Language Processing crucially leverages social media for market insights. However, most existing methods for this purpose rely on simple sentiment analysis models, which fail to capture the concrete trading intentions expressed in these discussions. While Large Language Models (LLMs) offer a promising alternative to simplistic sentiment analysis, the actual benefits of fine-tuning across different model families remain unclear in noisy, domain-specific contexts like online forums. To address this gap, we present a comprehensive assessment of the advantages and limitations of fine-tuning for Social Trading Action Detection (STAD), a novel task that aims to classify online posts into actionable categories, namely buy, sell, or other. In addition, we introduce FinReddit-2K, a manually annotated dataset consisting of 2123 Reddit posts, designed to serve as a benchmark for this task. Our experimental analysis goes beyond standard performance metrics and identifies both the types of errors that fine-tuning can successfully mitigate and those that it may inadvertently introduce. Through a systematic evaluation of 57 models, comparing 14 traditional models with 23 zero-shot LLMs and 20 fine-tuned variants, our results show that fine-tuning yields an average F1-score improvement of ＋15.1%. The best-performing model, a fine-tuned Mistral-7B, achieves an F1-score of 86.0%, although our analysis reveals that fine-tuning fails to produce meaningful performance gains in several scenarios.

Managing large public events involves significant challenges, including transportation congestion, attendee coordination, and enhancing the overall event experience. This paper introduces a novel approach that integrates Large Language Models (LLMs) and generative AI to address these issues effectively. We present a framework that leverages enriched knowledge graphs comprising diverse datasets, such as geographic information, transportation systems, environmental factors, and more, to recommend personalised, context-aware itineraries. By employing LLMs, we aim to manage attendee flow, stagger departure times, and guide individuals through engaging points of interest. Additionally, we incorporate generative AI to design gamified content, such as interactive quizzes and puzzles, tailored to user preferences. These gamification elements not only provide entertainment but also encourage staggered event departures, mitigating post-event congestion. The experimental study conducted in Milan demonstrated the effectiveness of the proposed system: AI-generated itineraries closely matched expected travel times, with minimal deviations of 2–5 minutes. Moreover, responses to the user experience questionnaire reflected high levels of usability, engagement, and overall satisfaction, reinforcing the potential of this approach for improving post-event mobility and attendee experience.

This paper explores the use of Large Language Models (LLMs) in mental health to assist psychologists and psychiatrists with diagnostic decision-making according to the ICD-11 classification system. ICD-11 is the 11th revision of the International Classification of Diseases, a globally used diagnostic tool for health conditions, including mental, behavioural, and neurodevelopmental disorders . In detail, we propose LLMind Chat, an AI-powered tool with a user-friendly interface designed to support mental health professionals in their diagnostic processes . LLMind Chat leverages a Retrieval Augmented Generation (RAG) model based on the Gemma 2 (27B parameters), specifically adapted to the context of the ICD-11. This RAG model combines the strengths of Gemma 2 with a comprehensive knowledge base derived from the ICD-11, allowing it to access and process relevant information from the classification manual in real-time. LLMind’s diagnostic accuracy was rigorously evaluated against the DSM-5-TR Clinical Cases manual, using automated metrics and mental health professionals’ expert validation. The result suggests that LLMind Chat can serve as a reliable decision-support tool, enhancing diagnostic reasoning and potentially reducing misclassifications.

Abstract Processing large-scale and highly interconnected Knowledge Graphs (KG) is becoming crucial for many applications such as recommender systems, question answering, etc. Profiling approaches have been proposed to summarize large KGs with the aim to produce concise and meaningful representation so that they can be easily managed. However, constructing profiles and calculating several statistics such as cardinality descriptors or inferences are resource expensive. In this paper, we present ABSTAT-HD, a highly distributed profiling tool that supports users in profiling and understanding big and complex knowledge graphs. We demonstrate the impact of the new architecture of ABSTAT-HD by presenting a set of experiments that show its scalability with respect to three dimensions of the data to be processed: size, complexity and workload. The experimentation shows that our profiling framework provides informative and concise profiles, and can process and manage very large KGs.

The problem of reliably predicting vehicle flows is paramount for traffic management, risk assessment, and public safety. It is a challenging problem as it is influenced by multiple factors, such as spatio-temporal dependencies with external factors (as events and weather conditions). In recent years, with the exponential data growth and technological advancement, deep learning has been adopted to approach urban mobility problems by addressing spatial dependency with convolutional neural networks and the temporal one with recurrent neural networks. We propose a spatio-temporal flow prediction framework, called 3D-CLoST, that exploits the synergy between 3D convolution and long short-term memory (LSTM) networks to jointly learn the characteristics of the space-time correlation from low to high levels. To the best of our knowledge, no method currently proposes such a structure for this problem. The results achieved on the two real-world datasets show that 3D-CLoST can learn behaviors from the data effectively.

We study the problem of group linkage: linking records that refer to multiple entities in the same group. Applications for group linkage include finding businesses in the same chain, finding social network users from the same organization, and so on. Group linkage faces new challenges compared to traditional entity resolution. First, although different members in the same group can share some similar global values of an attribute, they represent different entities so can also have distinct local values for the same or different attributes, requiring a high tolerance for value diversity. Second, we need to be able to distinguish local values from erroneous values.

The standardization and adoption of Semantic Web technologies has resulted in an unprecedented volume of data being published as Linked Data (LD). However, the “publish first, refine later” philosophy leads to various quality problems arising in the underlying data such as incompleteness, inconsistency and semantic ambiguities. In this article, we describe the current state of Data Quality in the Web of Data along with details of the three papers accepted for the International Journal on Semantic Web and Information Systems' (IJSWIS) Special Issue on Web Data Quality. Additionally, we identify new challenges that are specific to the Web of Data and provide insights into the current progress and future directions for each of those challenges.

In this paper we consider the case of the ICT adoption and use in an agriculture cluster in Lombardy, a northern region of Italy. At the state of the art, relationships among key factors of adoption and use of ICT in agriculture area received little attention by the academic literature. Thus, in this paper we aim to identify a research model in order to provide evidence of four different research questions concerning the determining factors for ICT adoption. The proposed case study reports and discusses the results obtained by analysing data from a survey of about 600 agricultural farms. Finally, Belief Bayesian Networks (BBNs) are used to analyse the complex influence relationships detected between research variables.

Many data sets containtemporal recordsover a long period of time; each record is associated with a time stamp and describes some aspects of a real-world entity at that particular time. From such data, users often wish to search for entities in a particular period and understand the history of one entity or all entities in the data set. A major challenge for enabling such search and exploration is to identify records that describe the same real-world entity over a long period of time; however, linking temporal records is hard given that the values that describe an entity can evolve over time (e.g., a person can move from one affiliation to another). We demonstrate the Chronos system which offers users the useful tool for finding real-world entities over time and understanding history of entities in the bibliography domain. The core of Chronos is a temporal record-linkage algorithm, which is tolerant to value evolution over time. Our algorithm can obtain an F-measure of over 0.9 in linking author records and fix errors made byDBLP. We show how Chronos allows users to explore the history of authors, and how it helps users understand our linkage results by comparing our results with those of existing systems, highlighting differences in the results, explaining our decisions to users, and answering "what-if" questions.

In medium-big enterprise it is quite typical that the database architecture is defined through a sequence of projects and realizations that result a number of different and sometime overlapping data sources. This trend is worsened by merger and acquisition activities that add in existing data architecture new data sources from external organizations. Data fragmentation reduces significantly the possibility to exploit organizational information assets and it needs the building of a data integration architecture. This architecture allows the organization to access to data stored by heterogeneous data sources and to manage updates through a unified view of this data. In this paper we present an original framework able to support the evolution of data architecture by identifying the optimal solution that maximize the quality of the overall architecture within a given cost threshold. In particular, we focus on Publish and Subscribe as solution for data architecture.

Web service evaluation is a phase of the Web service selection in which discovered Web services are evaluated with respect to user request, which means that the non functional properties (NFPs) offered by Web services are compared with the non functional properties requested by users. The fact that users and providers can express their NFPs in very flexible ways makes the management of NFPs a very complex task. In this paper we propose a computing-oriented description of NFPs and a novel approach to NFP-based service evaluation based on Hierarchical Constraint Logic Programming. This proposal extends our previous work on Policy Centered Meta-model (PCM).

In this paper we propose a meta-model for nonfunctional property descriptions targeted to support the selection of Web Services. The approach is based on the explicit distinction between NFP offered by providers and requested by users, on the concept of policy that aggregates NFP descriptions into single entities with an applicability condition, and finally on a set of constraint operators, which is particularly relevant for NFP requests. The semantic meta-model embracing the above perspective is defined by a BNF syntax whose semantics is formalized by an ontology. The ontology has been formalized in OWL-DL and WSML to provide for logical syntax. The logic upon which the meta-model supports NFP-based selection is discussed in the paper.

The discovery of a semantic web service (SWS) is the act of locating a machine-processable description of a SWS-related resource that may have been previously unknown and that meets certain functional criteria. The increasing availability of services that offer similar functionalities requires the discovery process to be enhanced with a selection phase that considers non-functional properties (NFPs) of services. This paper proposes a model to describe these properties and a novel approach to service selection. Our approach is based on the design of matching rules by means of mediators defined by sets of rules stating the condition for successful matches. These rules are based on the ontological description of objects representing NFPs that are required and offered. In particular, we define a set of rule schemas to support mediation and matching for a class of user-defined NFP-constraints clustered according to specified constraint operators. Rules support matching for both qualitative and quantitative non-functional properties.

Abstract This paper introduces a model‐driven approach to the design of collaborative Web‐based applications, i.e. applications in which several users play different roles, in a collaborative way, to pursue a specific goal. The paper illustrates a conference management application (CMA), whose main requirements include: (i) the management of users profiles and access rights based on the role played by users during the conference life cycle; (ii) the delivery of information and services to individual users; (iii) the management of the sequence of activities that lead to the achievement of a common goal. The presented approach is based on WebML, a conceptual modelling language for the Web. The paper also highlights some general properties—as understood by the practical experience of CMA development—that a Web modelling language should feature in order to fully support the development of collaborative applications. Copyright © 2003 John Wiley & Sons, Ltd.