| Summary: | 碩士 === 國立屏東科技大學 === 資訊管理系 === 90 === Search engines are useful tools in looking for information from the Internet. However, due to the methods of keyword-based similarity ranking used in the conventional search engines, people may still suffer from spending a large amount of time and effort to examine and browse the results presented by a search engine but not find the specific information he really wants. To remedy the above problem, in this thesis we develop a new search mechanism that can remove unrelated pages from Web search results. Through interacting with the users, the system can recommend them Web pages related their interests and extract critical features from the pages a user has selected. In addition, the Evolution Strategies Algorithm is employed to evolve the features to best describe the concept in a user’s mind. Therefore, the quality of Web search can be largely improved. By repeating the above procedure, new queries can be formulated for the search engines to retrieve more pages to match a user’s interests. Our experimental results show that the Evolution Strategies Algorithm is efficient and useful to enhance the quality of Web search.
|
|---|
