A Large Interlinked Knowledge Graph of the Italian Cultural Heritage

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 6280–6289 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 6280 A Large Interlinked Knowledge Graph of the Italian Cultural Heritage Stefano Faralli, Andrea Lenzi, Paola Velardi Department of Computer Science University of Rome Sapienza Rome, Italy {faralli, lenzi, velardi}@di.uniroma1.it Abstract Knowledge is the lifeblood for a plethora of applications such as search, recommender systems and natural language understanding. Thanks to the efforts in the fields of Semantic Web and Linked Open Data a growing number of interlinked knowledge bases are supporting the development of advanced knowledge-based applications. Unfortunately, for a large number of domain-specific applications, these knowledge bases are unavailable. In this paper, we present a resource consisting of a large knowledge graph linking the Italian cultural heritage entities (defined in the ArCo ontology) with the concepts defined on well-known knowledge bases (i.e., DBpedia and the Getty GVP ontology). We describe the methodologies adopted for the semi-automatic resource creation and provide an in-depth analysis of the resulting interlinked graph. Keywords: Cultural Heritage, Linked Open Data, Knowledge Graph, Ontology, ArCo, GVP, DBpedia 1. Introduction Numerous efforts have been made in recent years to create knowledge bases (Abu-Salih, 2021). Humans and machines require a formal definition of knowledge to enhance a variety of activities in support of appli- cations such as automated reasoning, cataloging, docu- mentation, discovery, promotion, and recommendation of large collections of entities and facts. Specifically, all the applications involving Machine Learning, or in general, Artificial Intelligence algo- rithms, perform better under the presence of large col- lections of data (Feigenbaum, 1984). Unfortunately, the creation of formal representations of knowledge is both a complex task, requiring the participation of ex- perts, and a time-consuming process. To address this problem, the Semantic Web and Linked Open Data (LOD) communities have provided mecha- nisms for the definition of formal ontologies and the interconnection of entities through knowledge bases (Gandon, 2018), thus helping both humans and ma- chines during the creation and engineering of knowl- edge bases. LOD data have been created in many do- mains, but in many others, they are still largely unavail- able. For example, despite the great interest in cultural heritage data (Mountantonakis and Tzitzikas, 2019), interlinking efforts are limited to relatively small re- sources, such as entities from libraries, archeologi- cal sites and museums (Pontes et al., 2020; Kyriaki- Manessi et al., 2018). In this paper, we describe the adopted methodologies and the resulting edition of a very large resource ob- tained within the ArCo, GVP and DBpedia Linking Initiative (AGDLI) (Faralli et al., 2021). AGDLI is a research activity within the project SMARTOUR: in- telligent platforms for tourism, funded by the Italian Ministry of University and Research, aimed at linking the ArCo’s catalogue of Italian cultural entities (Car- riero et al., 2019) to the well-known Getty Vocabulary Program (GVP) (Harpring, 2010) and DBpedia (Auer et al., 2007) ontologies, with the main goal of provid- ing a semantically rich and …

Content truncated to 500 words