Search

Data enrichment facilitates the creation of rich, expressive, and high-quality datasets, enabling valuable analytics and enhanced decision-making. The accurate geolocation of residential addresses and travel routes is crucial for determining the most appropriate locations of critical social infrastructure, such as educational and medical centres. This paper introduces an interactive semantic enrichment approach that enhances urban data by integrating high-quality geospatial information. The approach is supported by a modular and extensible data enrichment framework, which leverages existing geolocation services, enabling seamless data integration. Human-in-the-loop revision is employed to enhance the quality of geocoding results. A real-world pilot study conducted in Sofia, Bulgaria, was used to validate this approach, demonstrating its promising potential in addressing pressing issues in parametric urban planning.

La Interpretazione Semantica delle Tabelle (STI) è uno dei metodi più ampiamente utilizzati per identificare entità nei dati tabellari. In questo lavoro, viene delineata una metodologia per implementare il collegamento delle entità su larga scala utilizzando tecniche di apprendimento automatico, concentrandosi sulle sfide e le soluzioni associate alla gestione di grandi quantità di dati. La metodologia si basa sul concetto di corrispondenza tra tabelle e Knowledge Graph (KG), che è un passo fondamentale per arricchire ed estendere Knowledge Graph (KG) dai dati semistrutturati. Inoltre, sono stati affrontati i dettagli di STI, EL e del panorama emergente dei Large Language Models (LLM) e del loro potenziale applicazione in STI e EL. Inoltre, è stato dimostrato l'impatto dell'utilizzo di distinti Knowledge Graphs, come Wikidata e Wikipedia, nel contesto di EL utilizzando ChatGPT 4. In particolare, è esplorato il ruolo delle tecniche Human-in-the-Loop (HITL) nel migliorare le prestazioni del modello. Questo articolo delinea le basi fondamentali per lo sviluppo di un approccio scalabile che adatta il framework esistente acsti per gestire scenari di dati su larga scala, migliorando così la sua efficienza operativa e la sua applicabilità., Semantic Table Interpretation (STI) is one of the most widely used methods for identifying entities in tabular data. In this work, a methodology is delineated for implementing entity linking on a large scale using machine learning techniques, focusing on the challenges and solutions associated with handling vast amounts of data. The methodology is based on the concept of table-to-Knowledge Graph (KG) matching, which is a key step to enrich and extend Knowledge Graphs (KGs) from semi-structured data. Moreover, the intricacies of STI, EL, and the emerging landscape of Large Language Models (LLMs) and their potential application in STI and EL have also been touched upon. In addition, the impact of employing distinct KnowledgeGraphs, such as Wikidata and Wikipedia, in the context of EL using ChatGPT 4 has been demonstrated. In particular, the role of Human-in-the-Loop (HITL) techniques in enhancing model performance is explored. This paper outlines the foundational groundwork for the development of a scalable approach that adapts the existing acsti framework to handle large-scale data scenarios, thereby enhancing its operational efficiency and applicability.

This paper describes s-elBat, a Semantic Table Interpretation approach. The approach inherits and improves the part of the techniques belonging to the MantisTable, an approach used and tested in previous editions of the SemTab challenge. s-elBat adds an innovative and optimised lookup approach for generating candidate entities for the annotation.

When information available in unstructured or semi-structured formats, e.g., tables or texts, comes in, finding links between strings appearing in these sources and the entities they refer to in some background Knowledge Graphs (KGs) is a key step to integrate, enrich and extend the data and/or KGs. This Entity Linking task is usually decomposed into Entity Retrieval and Entity Disambiguation because of the large entity search space. This paper presents an Entity Retrieval service (LamAPI) and discusses the impact of different retrieval configurations, i.e., query and filtering strategies, on the retrieval of entities. The approach is to augment the search activity with extra information, like types, associated with the strings in the original datasets. The results have been empirically validated against public datasets.

In this paper, we present MammoTab, a dataset composed of 1M Wikipedia tables extracted from over 20M Wikipedia pages and annotated through Wikidata. The lack of this kind of datasets in the state- of-the-art makes MammoTab a good resource for testing and training Semantic Table Interpretation approaches. The dataset has been designed to cover several key challenges, such as disambiguation, homonymy, and NIL-mentions. The dataset has been evaluated using MTab, one of the best approaches of the SemTab challenge.

Annotating tables with high-quality semantic labels and linking cells in these tables to entities in reference knowledge graphs (KGs) serve several downstream applications such as knowledge graph construction, data enrichment, data analytics, and so on. There is an increasing interest in combining algorithms for the automatic annotation of tables with interactive applications that let users control and improve the annotations and use them for data enrichment. In this paper, we propose a supervised entity linking network that behaves as a re-ranker and score normalizer of scores returned by off-the-shelves entity retrieval systems. By combining these refined scores with additional signals of reliability of the ranking obtained thereof, we can estimate a confidence score, which tells how reliable is the decision to link a cell to the top-ranked candidate entity. The confidence score is used to decide whether to establish a link or not and to prioritize the cells to be revised by the users to improve the annotation quality. Experiments suggest that the proposed approach provides an effective approach to identifying most critical decisions and supporting link revision within a human-in-the-Ioop data enrichment paradigm.

In this paper, we present MantisTable V, a novel unsupervised and automatic approach for the Semantic Table Interpretation. The approach is performed against DBpedia and Wikidata, and it can be easily adapted to any other Knowledge Graph. Moreover, we provide a tool (LamAPI) that allows to efficiently fetch data needed for Semantic Table Interpretation tasks from the Knowledge Graph dumps. The approach is manageable through a User Interface (tUI), a separated tool, which allows the visualisation and modification of table data and semantic annotations.

Much information is conveyed within tables, which can be semantically annotated by humans or (semi)automatic approaches. Nevertheless, many applications cannot take full advantage of semantic annotations because of the low quality. A few methodologies exist for the quality assessment of semantic annotation of tabular data, but they do not automatically assess the quality as a multidimensional concept through different quality dimensions. The quality dimensions are implemented in STILTool 2, a web application to automate the quality assessment of the annotations. The evaluation is carried out by comparing the quality of semantic annotations with gold standards. The work presented here has been applied to at least three use cases. The results show that our approach can give us hints about the quality issues and how to address them.

In this paper, we present a novel unsupervised and automatic approach for Semantic Table Interpretation. The technique presented is performed against DBpedia and Wikidata, and it can be easily adapted to any other Knowledge Graph (KG). Moreover, we provide a tool (LamAPI) that allows to efficiently fetch data needed for Semantic Table Interpretation (STI) tasks from the KG dumps.

In this paper, we present MammoTab, a dataset composed of 1M Wikipedia tables extracted from over 20M Wikipedia pages and annotated through Wikidata. The lack of this kind of datasets in the state- of-the-art makes MammoTab a good resource for testing and training Semantic Table Interpretation approaches. The dataset has been designed to cover several key challenges, such as disambiguation, homonymy, and NIL-mentions. The dataset has been evaluated using MTab, one of the best approaches of the SemTab challenge.

This paper describes STILTool, an open-source tool for the automatic evaluation of the quality of semantic annotations computed by semantic table interpretation approaches. STILTool provides a graphical interface allowing users to analyse the correctness of the annotations of tabular data. The tool also provides a set of statistics in order to identify the most common error patterns.