Research On User Behavior Learning Based Agricultural Information Meta-search Engine

The information explosion makes more and more useful information drowning into information noise, so how to extract the needed information and take advantage of it is the key to support the steady development of a region or an industry. Even though the development of search engine has provided a more effective way for this situation, with the rapid growth of the amount of information, the normal search engine shows the negative signs, like the low information coverage, the low correlation between searching intent and results, and so on. In response to this problem, on one hand, the subject-oriented search engine is coming up, aiming to achieve an accurate information search in a small range of specific information resources; on the other hand, meta-search engine has had a significant development, seeking to integrate the results from other search engines and with a deep processing to achieve the comprehensive search results and the required accuracy.Our agricultural information development is also carried out rapidly. Significant increase of the agricultural sites number makes the agricultural information more abundant, so how to help the agricultural information user to dig out the needed information is the bottleneck restricting the development of our agricultural information. Several agricultural search engines have been developed and put into operation, but, in terms of current service quality and operating conditions, they are still not very good to meet the needs of users of agricultural information. The low information coverage leads to less user groups, which makes itself lacking the fundamental driving force for further development.Through the analysis of domestic and international mainstream search engines, meta-search engines and agricultural information oriented search engine, this paper learned the principles and technology of the existing search engine, understood their application status and identified their deficiencies. On this basis, this paper presents an agricultural information-oriented meta-search engine, with a detailed description to its framework and operation process. Meanwhile, this paper, for the shortcomings of existing search engine technology, presents a user behavior learning-based method. By studying user behavior on searching, it could have a dynamic deeper processing for results, in order to provide better search services for the users of agricultural information.On one hand, the meta-search system can integrate the results from both the comprehensive search engine and professional search engine, improving the coverage of agricultural information greatly so to attract a wide range of agricultural information user groups. On the other hand, through the real-time recording, feedback and learning of various acts when users search for information, the user behavior learning-based method can dynamically adjust the order of results sequence and the content and order of categories sequence, showing the most effective results with the cluster label at the most accessible places, greatly enhancing the effectiveness of the results and the clustering validity of results, giving the user with excellent search experience.In the end, prototype system in accordance with the concept of the proposed method confirmed the effectiveness and efficiency of the proposed agricultural information-oriented meta-search system, as well as the feasibility of user behavior learning-based method.