Learning Embeddings to lexicalise RDF Properties

Proceedings of the Fifth Joint Conference on Lexical and Computational Semantics (*SEM 2016), pages 219–228, Berlin, Germany, August 11-12, 2016. Learning Embeddings to lexicalise RDF Properties Laura Perez-Beltrachini and Claire Gardent CNRS, LORIA, UMR 7503 Vandoeuvre-l`es-Nancy F-54500, France {laura.perez, claire.gardent}@loria.fr Abstract A difficult task when generating text from knowledge bases (KB) consists in finding appropriate lexicalisations for KB sym- bols. We present an approach for lexicalis- ing knowledge base relations and apply it to DBPedia data. Our model learns low- dimensional embeddings of words and RDF resources and uses these represen- tations to score RDF properties against candidate lexicalisations. Training our model using (i) pairs of RDF triples and automatically generated verbalisations of these triples and (ii) pairs of paraphrases extracted from various resources, yields competitive results on DBPedia data. 1 Introduction In recent years, work on the Semantic Web has led to the publication of large scale datasets in the so- called Linked Data framework such as for instance DBPedia or Yago. However, as shown in (Rec- tor et al., 2004), the basic standards (e.g., RDF, OWL) established by the Semantic Web commu- nity for representing data and ontologies are diffi- cult for human beings to use and understand. With the development of the semantic web and the rapid increase of Linked Data, there is consequently a growing need in the semantic web community for technologies that give humans easy access to the machine-oriented Web of data. Because it maps data to text, Natural Language Generation (NLG) provides a natural means for presenting this data in an organized, coherent and accessible way. It can be used to display the content of linked data or of knowledge bases to lay users; to generate explanations, descriptions and summaries from DBPedia or from knowledge bases; to guide the user in formulating knowledge base queries; and to provide ways for cultural her- itage institutions such as museums and libraries to present information about their holdings in multi- ple textual forms. In this paper, we focus on an important sub- task of generation from RDF data namely lexical- isation of RDF properties. Given a property, our goal is to map this property to a set of possible lexicalisations. For instance, given the property HASWONPRIZE, our goal is to automatically in- fer lexicalisations such as was honored with and received. Our approach is based on learning low- dimensional vector embeddings of words and of KB triples so that representations of triples and their corresponding lexicalisations end up being similar in the embedding space. Using these em- beddings, we can then assess the similarity be- tween a property and a set of candidate lexicali- sations by simply applying the dot product to their vector embeddings. One difficulty when lexicalising RDF proper- ties is that, while in some cases, there is a direct and simple relation between the name of a prop- erty and its verbalisation (e.g., BIRTHDATE / “was born on”), in other cases, the relation is either in- direct (e.g., ROUTEEND / “finishes at”) or …

Content truncated to 500 words