Among data mining technique, clustering is one of the most important and traditional concept also an unsupervised learning paradigm. Similarity of a document pairs can be measured by matching of concepts. Finding or extracting the most relevant concept from the documents is a challengeable task. In this paper, the authors introduce a concept of multi view point based similarity measure. Their proposed methods uses multiple point of reference between document pairs to extract more relevant match concept rather than extracting only ideas based on similarity measure.