Though existing search engines have provided efficient tool for network information retrieval, there are still some problems. There are several improvement ways, such as eliminating unrelated documents or changing their order by deep analyzing for results of single search engine, combining or intersecting the results of multiple search engine to improve the recall or accuracy, which all limited for no local index bases. We decide to build up a new inteligent network retrieval system with our own local index bases .Constructing indopendently will cost much and too difficult to keep the integrity, to take advantages of search engine, we construct it on base of search engine. First, driven by a big keywords table, fetch URL from special search engine by CGI programming, then get all responding full documents and analyze them deeply to form the features (Second-hand Document, SD) table. After this, build up kinds of index base needed by possible retrieval requirement. During the information retrieval service, system will renew the bases to reflect real network information by automatic searching according to the old network structure, which was built by existing URL aggregation. The last perfect retrieval system is still not implemented, but hte initial performance is satisfying.
展开▼