The integration of databases and machine learning promises to enhance various aspects of data management, analysis, and application.However, in-database machine learning (In-DB ML) is not easily portable to different database management systems and current approaches are typically limited to training and inference, while modern machine learning pipelines often involve aspects such as continuous learning, unlearning, and explainability.This paper presents BornSQL, a In-DB ML algorithm based on the Born Classifier [7], and exclusively implemented through standard SQL queries.BornSQL can handle categorical data, and it is particularly appropriate for classification of textual data.Further contributions of BornSQL are i) incremental learning to efficiently enforce model updates when new data become available in the db, ii) unlearning when selected data needs to be excluded due to privacy issues, and iii) global/local explainability to associate the importance of a feature/attribute in determining the classification result.We illustrate the usage and scalability of the algorithm using a benchmark database consisting of 2,359,828 scientific publications divided into three classes and composed of 3,942,559 features.The training time is linear in the number of publications and the average inference time for a publication is 1 millisecond on our experimental environment.We discuss potential applications such as cost-effective model serving, exploratory data analysis, and data privacy.

Despite recent progress, Biomedical Entity Linking (BEL) with large language models (LLMs) remains computationally inefficient and challenging to deploy in practical settings. In this work, we demonstrate that instruction-tuning of open-source generative models can offer an effective solution when applied at the re-ranking stage of the BEL pipeline. We propose a set-wise instruction-tuning formulation that enables fast and accurate candidate selection. Our method demonstrates strong performance on multiple BEL benchmarks, yielding significant improvements in linking accuracy (3%–24%) while reducing inference time compared to the state-of-the-art. We integrate our generative re-ranker into BeLink, a modular, end-to-end system designed for practical real-world BEL applications.

Abstract Modern data mining applications require to perform incremental clustering over dynamic datasets by tracing temporal changes over the resulting clusters. In this paper, we propose A-Posteriori affinity Propagation (APP), an incremental extension of affinity propagation (AP) based on cluster consolidation and cluster stratification to achieve faithfulness and forgetfulness. APP enforces incremental clustering where i) new arriving objects are dynamically consolidated into previous clusters without the need to re-execute clustering over the entire dataset of objects, and ii) a faithful sequence of clustering results is produced and maintained over time, while allowing to forget obsolete clusters with decremental learning functionalities. Four popular labeled datasets are used to test the performance of APP with respect to benchmark clustering performances obtained by conventional AP and incremental affinity propagation based on nearest neighbor assignment algorithms. Experimental results show that APP achieves comparable clustering performance while enforcing scalability at the same time.

We present Canvas Conversation Tales (CCT), an edutainment web application for creating imaginary dialogues about paintings. To encourage participation, CCT employs techniques like introductory onboarding, possible story branching, expression of story preferences, and chat among users. This demo invites users to explore two usage scenarios of CCT, namely the development of a new story, and the creation of a new branch based on an existing story developed by others. We expect the first scenario to engage users open to express their creativity, and the second scenario - those willing to questions and discussions about the conversations of others.

Current approaches for detecting text reuse do not focus on recontextualization, i.e., how the new context(s) of a reused text differs from its original context(s).In this paper, we propose a novel framework called TRoTR that relies on the notion of topic relatedness for evaluating the diachronic change of context in which text is reused.TRoTR includes two NLP tasks: TRiC and TRaC.TRiC is designed to evaluate the topic relatedness between a pair of recontextualizations. TRaC is designed to evaluate the overall topic variation within a set of recontextualizations.We also provide a curated TRoTR benchmark of biblical text reuse, human-annotated with topic relatedness.The benchmark exhibits an inter-annotator agreement of .811.We evaluate multiple, established SBERT models on the TRoTR tasks and find that they exhibit greater sensitivity to textual similarity than topic relatedness.Our experiments show that fine-tuning these models can mitigate such a kind of sensitivity.

To cope with the growing volume, complexity, and articulation of legal documents as well as to foster digital justice and digital law, increasing effort is being devoted to legal knowledge extraction and digital transformation processes. In this paper, we present the ASKE (Automated System for Knowledge Extraction) approach to legal knowledge extraction, based on a combination of context-aware embedding models and zero-shot learning techniques into a three-phase extraction cycle, which is executed a number of times (called generations) to progressively extract concepts representative of the different meanings of terminology used in legal documents chunks. A graph-based data structure called ASKE Conceptual Graph is initially populated through a data preparation step, and it is continuously enriched at each ASKE generation with results of document chunk classification, new extracted terminology, and newly derived concepts. A quantitative evaluation of ASKE knowledge extraction and document classification is provided by considering the EurLex dataset. Furthermore, we present the results of applying ASKE to a real case-study of Italian case law decisions with qualitative feedback from legal experts in the framework of an ongoing national research project.

In this paper, we present CRIKE, a data-science approach to automatically detect concrete applications of legal abstract terms in case-law decisions. To this purpose, CRIKE relies on the use of the LATO ontology where legal abstract terms are properly formalized as concepts and relations among concepts. Using LATO, CRIKE aims at discovering how and where legal abstract terms are applied by judges in their legal argumentation. Moreover, we detect the terminology used in the text of case-law decisions to characterize concrete abstract-term instances. A case-study on a case-law decisions dataset provided by the Court of Milan, Italy, is also discussed.

In this paper, we present an approach for classifying heterogeneous Web information resources and organizing them as smart entity views based on a target entity of interest (e.g., person, place, town, event). In particular, we propose a dimensional clustering algorithm for the generation of smart entity views and we provide analysis techniques for enabling web resource exploration through entity visualization, entity analysis, and entity expansion.

The goal of this paper is to demonstrate inWalk, an interac-tive web-based system for linked data exploration featured by the notion of inCloud and thematic walk. The demon-stration focuses on the key functionalities of the system for smart data aggregation and navigation.

The huge amount of social data actually available and everyday produced demands for discovery techniques to mine prominent information topics. In this paper, we present topic-clouds as a solution for thematic, conceptual exploration of social data, with specific application to microblogging posts of Twitter. Topic-clouds are the result of a discovery approach based on classification and abstraction techniques to mine the most prominent topics that emerge from a possibly-large set of social data.

The development of solutions to effectively browse and explore the huge amount of data actually available in the Linked Data Cloud is getting more and more importance. As a result, automated techniques to generate high-level, concept-based information structures are recently being emerging. The notion of inCloud we proposed is an example in this direction. In this paper, we propose to extend the in Cloud construction process with crowdsourced knowledge, to benefit from human knowledge and experience for improving the overall quality of resulting inClouds for linked data exploration.

P2P techniques for semantic coordination based on semantic communities are recently emerging to enforce effective collaboration platforms. In this paper, we focus on semantic coordination in the iCoord P2P system and we introduce the notion of collective intelligence to enable semantic peer communities to setup a shared community vocabulary which is gradually built with the autonomous contribution of all the community members according to their terminological preferences. Techniques for extending iCoord to include in the collective intelligence also the social peer knowledge coming from conventional tagging systems are also presented.

The recent success of distributed and dynamic infrastructures for knowledge sharing has raised the need for semiautomatic/automatic ontology evolution strategies. Ontology evolution is generally defined as the timely adaptation of an ontology to changing requirements and the consistent propagation of changes to dependent artifacts. In this article, we present an ontology evolution approach in the context of multimedia interpretation. Ontology evolution in this context relies on the results obtained through reasoning for the interpretation of multimedia resources, through population of the ontology with new individuals or through enrichment of the ontology with new concepts and new semantic relations. The article analyses the results of interpretation, population and enrichment obtained in evaluation experiments in terms of measures such as precision and recall. The evaluation reveals encouraging results.

In this article, we describe the MX-Onto ontology for providing a Semantic Web compatible representation of music resources based on their context. The context representation is realized by means of an OWL ontology that describes music information and that defines rules and classes for a flexible genre classification. By flexible classification we mean that the proposed approach enables capturing the subjective interpretation of music genres by defining multiple membership relations between a music resource and the corresponding music genres, thus supporting context-based and proximity-based search of music resources.