Ontology Construction Based On Cluster-FCA-Merge Algorithm

Ontology is a clear shared conceptual model of formal specification and the core of the next generation Internet. It is significant and widely used in information management, knowledge mining and knowledge sharing. Ontology construction is a kind of automatic or semi-automatic ontology construction technology and it is the starting point for ontology application.Firstly, this paper introduces the background of the research and the related technology---document clustering.Secondly, it reviews the current situation and trend of ontology construction. The time and space efficiency of existing ontology construction algorithms are low and the quality of the output ontology is poor. According to "divide-and-conquer", the Cluster-FCA-Merge algorithm is proposed. Its procedure:(1) using the improved k-means clustering algorithm document to cluster. The improved k-means clustering algorithm is characterized by the starting point for clustering document is identified, rather than as the traditional k-means clustering as a random starting point selection. (2) According to each result of clustering, construct an ontology. (3) Calculate the distance between each body to merge two ontologies by using FCA-Merge algorithm if their distance is less than a threshold.Finally, it calculates the lexical recall and lexical overlap between the ontology and manually constructed ontology.