Research Of The Result Merging Of Meta Search Engine Based On Formal Concept Analysis

Meta Search Engine is a search tool which calls the other independent search engine to achieve the information retrieval. It usually doesn't have independent database and achieves the search task by selecting member search engines, collecting web pages and ranking all the results. So the scheduling strategy of member search engine and ranking aggregation of search results are the key technologies of meta search engine. Scheduling strategy ensures meta search engine select the proper member search engine to participate in the search task. It improves the efficiency of searching on one hand, on the other hand it helps to provide the better database for the rank aggregation of meta search engine. Merging the results of meta search engine is more important, because the results eventually reflect the accuracy, relevance and the performance of meta search engine. Therefore, a good scheduling strategy and ranking agrregation can not only effectively improves the search coverage of meta search engine, but also improves the accuracy and relevance of the search results. Based on these ideas, the paper carries through a research, the main points of the research as follows:1. Every member search engine has its own area of coverage on the internet. In order to take a better advantage of every member search engines, the paper put forward a new scheduling strategy of member search engine. In order to determine the search capability of every member search engine in specific areas, the paper uses static study method firstly. Meta search engine should make sure the area which the user's search query belongs to. And then by comparing the search capability of every member search engines in the area, meta search engine selects appropriate member search engines to participate in the search task. When the search ended, meta search engine should analyze user's feedback and adjust the search capability of member search engine to supervise the next search task.2. A result merging method based on formal concept analysis has been put forward in the paper. Make the member search engines which participate in the search task as the attribute set and make all the web pages returned by all the member search engines as the object set. Then a concept lattice will be built. Merge the results during the process of traversing the concept lattice. The merging result process can be viewed as an election process. Member search engine which participate in the search are voters and all the pages are candidates. This means that if a document has been retrieved by more member search engines, the more important the document is. If many documents were retrieved by the same number of member search engines, then the importance of these documents depends on the search ability of member search engines which retrieved them and their original sequence in member search engine.3. In this paper, a meta search engine system based on the scheduling strategy and the merging algorithm mentioned above has been built. The performance of the system has been proved by experiment and analysis. Given keywords, the paper compare the search result of several meta search engines and evaluate the performance of the system comprehensively.