Online information about flamenco music is scattered over different sites and knowledge bases. Unfortunately, there is no common repository that indexes all these data. In this work, information related to flamenco music is gathered from general knowledge bases (e.g., Wikipedia, DB- pedia), music encyclopedias (e.g., MusicBrainz), and specialized flamenco websites, and is then integrated into a new knowledge base called FlaBase. As resources from different data sources do not share common identifiers, a process of pair-wise entity resolution has been performed. FlaBase contains information about 1,174 artists, 76 palos (flamenco genres), 2,913 albums, 14,078 tracks, and 771 Andalusian locations. It is freely available in RDF and JSON formats. In addition, a method for entity recognition and disambiguation for FlaBase has been created. The system can recognize and disambiguate FlaBase entity refer- ences in Spanish texts with an f-measure value of 0.77. We applied it to biographical texts present in Flabase. By using the extracted information, the knowledge base is populated with relevant information and a semantic graph is created connecting the entities of FlaBase. Artists relevance is then computed over the graph and evaluated according to a flamenco expert criteria. Accuracy of results shows a high degree of quality and completeness of the knowledge base.

This work was funded by the COFLA2 research project (Proyectos de Excelencia de la Junta de Andaluca, FEDER P12-TIC-1362) and the SIGMUS research project (TIN2012- 36650).