A Clustering Algorithm Using Twitter User Biography

Our previous work proposed a clustering algorithm to cluster research documents automatically. It used Web hit counts of AND-search on two words as a document vector. Target documents are clustered with a result of k-means clustering method, in which cosine similarity is used to calculate a distance. This paper uses this algorithm to cluster twitter users. However, the twitter users have different characteristics from the research documents. Therefore, we investigate problems of the using our algorithm for twitter users and propose some ideas to resolve it.