Proposal of a recommendation method of Open Data Predicates using statistical processing

Abstract

　In recent years, many administrative organizations and local governments are engaged in opening and utilizing open data but until the data utilization has not reached where the government promotes "data can be secondarily used without much manpower being machine-readable". If these open data can be easily converted to RDF format by using common predicates, utilization of open data by linking data is promoted. In this research, we applied an application of automatic generation of predicates using Word2Vec algorithm to convert data disclosed as open data into RDF format. We propose a predicate recommendation method to convert predicates corresponding to item names by using statistical processing called “predicate vector method” which is vectorizing each data item of open data and converting it to RDF format.