Building semantic concept networks by wikipedia-based formal concept analysis

A concept network is a form of knowledge base that is mandatory for semantic search, personalized search, recommendation, and text mining. Recently, one of most feasible methods to build concept networks is to extend the representation of concepts with ‘world knowledge’-level ontology. In this paper, we introduce a novel way of building concept networks with the 3rd-order text tensor model using wikipages in Wikipedia ontology. Normally, the relationships among concepts tend to slowly change over time, and thus we intend to represent the concepts according to the principle of formal concept analysis (FCA). Basically, we define a Wikipedia article as a concept, which is represented as a ‘term-by-document’ matrix for the current document collection. Consequently, we can obtain more reasonable concept networks being aware of the current trends.