Linked Open Data has been recognized as a useful source of background knowledge for building content-based recommender systems. Vast amount of RDF data, covering multiple domains, has been published in freely accessible datasets. In this paper, we present an approach that uses language modeling approaches for unsupervised feature extraction from sequences of words, and adapts them to RDF graphs used for building content-based recommender system. We generate sequences by leveraging local information from graph sub-structures and learn latent numerical representations of entities in RDF graphs. Our evaluation on two datasets in the domain of movies and books shows that feature vector representations of general knowledge graphs such as DBpedia and Wikidata can be effectively used in content-based recommender systems.
RDF graph embeddings for content-based recommender systems
ROSATI, JESSICA;DE LEONE, Renato;
2016-01-01
Abstract
Linked Open Data has been recognized as a useful source of background knowledge for building content-based recommender systems. Vast amount of RDF data, covering multiple domains, has been published in freely accessible datasets. In this paper, we present an approach that uses language modeling approaches for unsupervised feature extraction from sequences of words, and adapts them to RDF graphs used for building content-based recommender system. We generate sequences by leveraging local information from graph sub-structures and learn latent numerical representations of entities in RDF graphs. Our evaluation on two datasets in the domain of movies and books shows that feature vector representations of general knowledge graphs such as DBpedia and Wikidata can be effectively used in content-based recommender systems.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.