Multi level and multi attribute content depth measure based web document clustering for efficient web search

Abstract

Users of web search engines are habitually forced to examine newlinethrough the long ordered list of documents reverted by the engines. The newlineInformation Retrieval (IR) community has surveyed document clustering as newlinean apparent method of organizing retrieval results. The huge increase in newlineamount of information present on web, positions new challenges in clustering newlineregarding the underlying data model and nature of clustering algorithm. Many newlineconventional document clustering techniques mostly rely on single term newlineanalysis of document data set. Clustering similar documents from a large collection is not a one step task. It involves multiple stages which generally comprise three main newlinephases: feature extraction and selection, document representation, and newlineclustering. To achieve more accurate document clustering, more informative newlinefeatures such as topic, structure, layout, visual, content, semantic are essential newlineto be extracted in order to achieve efficient clusters of Web documents. The newlinemajor key requirements for web document clustering methods are relevance, newlinebrowsable summaries, overlap, snippet tolerance, speed, scaling and so on. newlineThe aim of this thesis is to improve the efficiency and accuracy of web newlinedocument clustering by considering the features that are extracted along with newlinecontent depth of the document. newline newline