A search quality evaluation based on FCA

Abstract

Commercial search engines, especially meta-search engines was designed to retrieve the information by submitting users’ queries to multiple conventional search engines and integrating their partial searching results generated by each search engine. How to find the best search engine for user queries is know as the selection problem of search engines. Recently, the selection problem has become a research hotspot in meta-search engines. Therefore, it is the key technology to evaluate the quality of the search result of every search engine. In this paper, we proposed the search result’s quality evaluations by FCA. First we submit a query to seven public search engine (Google, yahoo, msn, soso, sogou, baidu, zhongsou), then with seven search results returned. Each search result composes with some urls and some snippets for the urls. As a url, different search engine provide its snippet is different often. Therefore, we submit the url to common spider. Analyse its content, and then get the keyword for the url. At last, each search engine result has two sets: one is the url set, the other is the keyword set. In addition, there is a relation ship between the url set and the keyword set (whether the url contains the keyword, or whether the keyword applies to the url).from the search result context, we can get seven search result concept lattice. We evaluate the search result by analysis the correlation among the seven-search result concept lattice.