The Research And Implementation Of Personalized Information Retrieval System Based On Osmot Search Engine

With the rapid development of Internet, information is growing explosively. Webis becoming an important way of getting information. The Internet informationretrieval platform, for example, search engine (Baidu and Google, etc) is faced withmany difficulties when being widely used. Firstly, for the same keywords, differentusers might have the completely different information requirements. Secondly, user’sinformation demand will change with time. While, the information retrievaltechnology that the existing search engines used is mainly unified mode (one size fitsall), the keyword match pattern. They didn’t consider the individual differencesbetween users and timeliness, thus affecting the accuracy of the retrieval results andthe user’s satisfaction. The emergence of personalized information retrieval systemprovides a way to solve these difficulties.At present, the important reason that the information retrieval system is difficultto implement personalized function, is that it can’t obtain the user’s retrieval needsaccurately. In many of the information needs, learning and research are relativelystable ones. This is because each user has its own major and the direction of research,these are stable, at least in quite a long period are stable. If the user’s informationneeds are stable, to implement individualized function in information retrieval systemis a relatively easy thing.In theory research field, there have had a lot of research results in personalized information retrieval area. For example: the user behavior research, optimizing searchengines using click through data. Machine learning research, the researchers designeda learning algorithm, used SVM in information retrieval system to improve theaccuracy of the information retrieval results. Based on these theoretical researches,researchers have developed search engine with personalized function, and opened thesource code, Osmot search engine is one of them.At first, this paper studied the Osmot search engine’s source code, and modifiedit. The IKAnalyzer was integrated into the Osmot search engine, this increased thefunction of intelligent Chinese word segmentation and made the Osmot search enginesupport Chinese searching. On this basis, combined with open source software andframework such as Lucene, Struts, iBatis, etc, the author designed and implementeda personalized information retrieval system. This system can record user’s searchhistory and analyze it, and run machine learning on it. At last it can rerank the originalsearch result by using the machine learning result. On the basis of realizing thissystem, this paper also ran a preliminary test on the effectiveness of learning andrerank function by using the real data.