Improved clustering technique using metadata for text mining

摘要

In many text mining applications, information from Document is present in the form of Text along with Side Information or Metadata. Examples of this side information include links to other web pages, title of the document, author name or date of Publication which are present in the text document. Such metadata may possess a lot of information for the clustering purposes. But this Side information may be sometimes noisy. Using such Side Information for producing clusters without filtering it, can result to bad quality of Clusters. So we use an efficient Feature Selection method to perform the mining process to select that Side Information which is useful for Clustering so as to maximize the advantages from using it. The proposed technique makes use of the process of Two-mode clustering which is a data mining technique that allows producing groups by Clustering both Text and Side Information.