A novel approach to aggregation of web documents by semantic suffix tree and self organizing feature map methods

The amount of information available on the internet became enormously high because of the rapid development in the electronic medium. Now days, managing and retrieving useful information from the web becomes the very tedious job. Web document clustering is a methodology which tries to group the available web documents in a meaningful way. There are lot of issues are identified in this area like, heterogeneous sources of information, no universal model, dynamicity of information, the huge volume of data, etc. most of the existing document clustering models follows the “bag-of-words” document representation. The proposed model presents an incremental analysis of web documents based on semantic suffix tree which analyses web documents by means of concepts to improve the cluster quality. The concepts are evaluated at three levels, a sentence level analysis, a document level analysis and domain level analysis. The proposed model also uses self organizing feature map (SOFM) to group similar documents together in a cluster and organize similar clusters close together unlike most other clustering methods. A set of experiments using the proposed model on different data sets have been conducted. The experiment result shows a significant improvement in the cluster quality.