Research On Desktop Based Domain Specific Meta-Search System

Search engines are one of the most useful tools on the Internet. Their main technical guidelines are coverage and accuracy. However,search engines' performances of today are quite poor when measured by these two standards. Statistics has shown that the most powerful search engine cover less then 40% of the Web's information. Even more,many other problems lie in search engines that cause the finding of wanted information a difficult task,e.g. low search accuracy,different information quality,different search interface,different level of personalized service. More investigations into search engines are requested,and mate-search systems emerged in the process. On one hand,meta-search engines improve coverage by invoking multiple search engines. On the other hand,they take measures to filter out useful information and reorganize them for the convenience of browsing. Main works of this thesis are as follows:(1) Reviewed the history of information searching on the Internet. Introduced the techniques concerned with search engines. Described problems remained in the field of search engines. Finally, a framework of DDSMS (Desktop Domain Specific Meta-Search) was present for discussion.(2) Based on the analysis of personal services on the Internet,we offered a simple and feasible way to improve personalizing results. It tracks user's entire searching process,collect information that could be used to determine his related characters and understand his search target better.(3) Through the comparison of different result binding methods of meta-search systems,we recommended the combination of clustering with other techniques. A suitable algorithm for on line meta-search result clustering was also suggested.