关键词: 个性化网络蜘蛛, 最优搜索算法, 在线增量自学习, 网页信噪比

Abstract: NonHogSearch, a topic-specific search engine based on improved bestfirst search algorithm was designed and implemented, which decreased the searching greed degree. Signal-to-noise ratio of Web page was used to judge whether or not the page belonged to the search topic. Further NonHogSearch Web spider made online-incremental adaptive learning, the reward generated directly by the ontopic pages would be feedback along the link-chain to update all the value Q of the links. NonHogSearch avoids going into local best solutions space earlier, and the performance of Web spider was improved. Experiments prove that it has better recall rate and precision rate than others.

Key words: topic-specific Web spider, best-first search algorithm, online-incremental adaptive learning, signal-to-noise ratio of web page