Tables are crucial containers of information, but understanding their meaning may be challenging. Over the years, there has been a surge in interest in data-driven approaches based on deep learning that have increasingly been combined with heuristic-based ones. In the last period, the advent of Large Language Models (LLMs) has led to a new category of approaches for table annotation. However, these approaches have not been consistently evaluated on a common ground, making evaluation and comparison difficult. This work uniquely compares Semantic Table Interpretation (STI) approaches with generative and encoder-only LLMs on diverse datasets. In particular, we conduct an extensive evaluation of four STI state-of-the-art (SOTA) approaches — Alligator (formerly s-elBat ), TURL , TableLlama , and DAGOBAH (the latter with a partial evaluation due to its high computational demands); Alligator and DAGOBAH belong to the family of heuristic-based algorithms, while TURL and TableLlama are respectively encoder-only and decoder-only LLMs. We also include in the evaluation both GPT-4o and GPT-4o-mini , since they excel in various public benchmarks. The primary objective is to measure the ability of these approaches to solve the entity disambiguation task concerning both the performance achieved on a common-ground evaluation setting and the computational and cost requirements involved, either monetary or in terms of computational resources, with the ultimate aim of charting new research paths in the field.

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.

This paper presents PANTHER, a neural network model for automatic handwriting extraction and recognition in psychodiagnostic questionnaires. Psychodiagnostic tools are essential for assessing and monitoring mental health conditions, but they often rely on pen-and-paper administration, which poses several challenges for data collection and analysis. PANTHER aims to address this problem by using a convolutional neural network to classify scanned questionnaires into their respective types and extract the patient’s responses from the handwritten annotations. The model is trained and evaluated on a dataset of five questionnaires commonly used in psychological and psychiatric settings, achieving high accuracy and similarity scores. The paper also describes the creation of an open-source library based on PANTHER, which can be integrated into a digital platform for delivering psychological services. This paper contributes to the field of computer vision and psychological assessment by providin...

Recently, an increasing interest has been in extracting and annotating tables on the Web. This activity allows the transformation of textual data into machine-readable formats to enable the execution of various artificial intelligence tasks, e.g., semantic search and dataset extension. Semantic Table Interpretation (STI) is the process of annotating elements in a table. The paper explores Semantic Table Interpretation, addressing the challenges of Entity Retrieval and Entity Disambiguation in the context of Knowledge Graphs (KGs). It introduces LamAPI, an Information Retrieval system with string/type-based filtering and s-elBat, an Entity Disambiguation technique that combines heuristic and ML-based approaches. By applying the acquired know-how in the field and extracting algorithms, techniques and components from our previous STI approaches and the state of the art, we have created a new platform capable of annotating any tabular data, ensuring a high level of quality.