|
For too much general search engine search results , and the theme is not strong , the subject-oriented search engine as the core research and public opinion information design theme search reptiles , thematic search engine the study has laid a good foundation . The paper first describes the background of the subject-oriented search engine , then the search engine on the theme of public opinion needs analysis system based on the outline design and detailed design , done some research on the system implementation and testing . Search reptiles collected as a subject-oriented public opinion , a selection policy based on the Boltzmann action online incremental self - learning ability , subject-oriented search crawler prototype selection strategy , based on the Boltzmann action by the greed of control network spider to crawl . This paper introduces the concept of signal-to-noise ratio of the page , the measure of the keywords in the web and through it exceeds a certain threshold , in order to determine the priority of web search topic pages , whether to adjust the Web search . On this basis , subject-oriented search crawlers introduction of the crawling process automatically update the learning model , when the topic page return , will return along the Links link feedback to update the Q value of all the links on the link . By doing this , the network spiders to avoid premature fall into the trap of local optimum subspace in the Web search space , and can accurately identify the topic of the page , with good self - learning ability . According to the value of the links in the link - priority queue size comparison Web Search priority . Based on these studies , design and subject-oriented search reptiles , and through comparison experiment with other search strategies , confirmed that the proposed algorithm has a certain superiority in the recall rate and precision rate in finding optimal action sequence has advantages . However, the mechanisms of how to improve the judgment of the weights of the current page , so the calculation of the value of network links more accurate , more intelligent search for reptiles , need further study later .
|