Research On Vertical Search Technology In Agriculture

With the explosive growth of network information,it has become more and more difficult for people to query the information they need.The emergence of search engines to some extent alleviated this contradiction.General search engines, which are now widely used, provide services for users at all kinds of fileds and in all different places. Although they easily achieved the relevant search,their narrow coverage and low retrieval efficiency made themself powerless.For changing this situation, some of them has changed the way of collecting comprehensive information originally into collecting the professtional informaiton in one field, thus providing a precise query results for users.If the relatively comprehensive and accurate search results are wanted, We must change the traditional ways of searching.Vertical Search Technology in Agriculture is generated against in this background. However, on the search based on keywords, the same keywords in different fields have different meanings, which bring out the low precision of the general search engine. Because the general search engines can not identify field users'true intentions. Agricultural field, as the thematic field, also have a lot of keywords with field significances, such as Fuji,corn and so on. For search users in the field,fuji is correlated with apples, not fuji camera nor Fujiyama.Corn is agricultural product, not Li Yuchun's fans.The aim of the research on vertical search technology in agriculture is to make the search engine understand the users'real intentions and feedback the final results set to the users. Firstly, the author built a meta-search engine on studying selection strategy and results treating methods of general search engine, analyzed how to use regular expressions to crawl the information within the web page in particular field, and then built a web feature library by field feature words and domain analysis method, and finally filtered results of the meta-search egine using above library. The vertical technology is completed with the combination of these three parts. The research is as follows:Firstly, the author focused on the struture of meta-search engine and the key technologies of the realization, which are selection strategy of member search engine and results treating methods, on the base of analysis of relevant documents, as well as classification and key technologies of genereal search engine. On selection strategy, the author analized, comparised and evaluated the existing methods, and then drew up a suitable strategy for the paper. Besides, in searching effects, the feedback results from the member search engines always wanted to be able to get closer to the optimal results, while the results of member search engines themselves were always unstructured. So the results processing is another key. point of this paper. During the processing of the results, the paper decribed, analized and comparized some common methods of results processing, and then concluded the results processing method appropriated for the system.Secondly, construction of the domain feature library of web page was for the service of filtering results set that the meta-search engine got initially.In accordance with this thought, it excluded non-domain data, reranged the resluts set and improved the accuracy of the results set. So the construction of domain feature library is another key point of the system. There are two basic elements in the construction. One is how to use regular expression to crawl information in the web page and the guarantee for the efficiency of the crawl. Another is that, after crawling the informaiton on the web page successfully, how to reuse regular expressions and domain analysis methods to analyze the information, and finally complete the construction of the library.Finally, the system discussed how to use the above library to filter the results sets which the meta-search engine got initially and reranged it, and then make the results sets back to the users.