Multi Level Web Data Extraction Based Topical Visual Structure Clustering for Efficient Web Search

摘要

The problem of web clustering has been approached in various strategies; however they suffer to achieve performance due to the poor data extraction and clustering approaches used. The most methods do not use all the features of web document than the textual features. To improve theperformance of web data extraction and clustering to support web search, the author present an efficient multi level web data extraction technique in this paper. The web document has been preprocessed to obtain various features from the text, structural and visual features. Extracted featureshave been used to perform Topical-Visual-Structure Clustering. Each class of the cluster has been organized into three different sub classes. The method first computes the topical similarity measure to identify the class of the document. Then the visual similarity and structural similaritymeasure has been used to identify the next level subclass of each cluster. The method focused to improve the performance of the web search and from the input query the method identifies the type of result the user expects. The proposed method increases the performance of web data extractionand web search.