Personalized Push Model And System Of Agricultural Science And Technology Information Service

With the rapid development of Internet and the popularity of personal computers,agricultural science and technology information can be found anywhere in Internet. However,since farmers know little about information technology, they have no ability to quickly accessthe right information they want, which finally hinders the propagation of agricultural scienceand technology information and helps incur the phenomenon of Information Island. To solveproblems concerning the low sharing rate and decentralization of such information, this papertakes advantage of authorized web sites as its information sources and focus on analyzing theweb logs and web pages. This paper completed the following tasks:(1) The research on methods to dynamically obtain agricultural science and technologyinformation. In order to make some help in solving the problem of asymmetric distribution ofsuch information, the focused crawler technology is used to get theme related WebPages;Secondly, the wrapper technique is utilized to extract agricultural attribute information andthen the regular expressions are applied to match structural information with the unstructureddata, and then the processed information is put into database; Thirdly, in order to improve thesuccess rate of extracting, task scheduling mechanism is used to divide the extracted task intosmall segments and failures retry mechanism is realized by using the user-defined xml file;Finally, experimental results are given to show the information extraction F-index is around85%. And the success ratio is improved by more than10%after the using of failures retrymechanism.(2) The study on personalized push model of agricultural science and technologyinformation service. In order to solve this problem that farmers who are unable to accuratelylocate the right information interested them easily, A representative user navigation pattern isestablished after a series of processes which are preprocessor of the web log file,vectorization of user sessions, and classification of user sessions by using improved k-meansclustering method; key contents of WebPages are extracted and the sets of representativefeature words are obtained by using Chinese segments technology. Then the TF-IDF methodis used to calculate each words’ weight to construct user profiles. Finally, visitor’s future requests are predicted by calculating the different values between visitors’ profiles and thetypical user profiles, which contribute to implement personalized information push service.Experimental figures show the predictive accuracy of the proposed personalized informationpush service model gained by the above method is69.23%compared with its recommendedaccuracy which is43.34%.(3) After a systematical analysis on both the extraction method and the method used forpersonalized push model, the object-oriented programming is used to design and realize thefinal system which provides users with lots of convenience such as dynamically obtainingagricultural science and technology information, personalized information pushing. In a word,this paper prompts the share and propagation of agricultural science and technologyinformation. Software tests show the system is available, robust and maintainable.