Supplement keyword-based indexing and models based on linkage information (like PageRank algorithm), with data mining techniques for web click stream analysis for ranking the web pages. This comprised 3 main areas:
-
Web content mining: Determining content relevance. This involved determining frequency, relative distance(among query words) and location of query words on a given page.
-
Web structure mining: Determining imporatance of a web page based on linkage information(Page Rank Algorithm). This involved ranking each page based on which web pages point/link to this page; and what does the link-text say about the web page.
-
Web usage mining: Using user’s search pattern, trend to improve search results. This involved learning NN-model through user clicks.