The Study Of User-Oriented Information Retrieval System

Obtaining interested information through the Internet is one of users’ most basic needs in using Internet (others included entertainment, social and so on). According to the rank of 2008 released by well-known Media Research agency Nielsen Inc. and Alexa (www.alexa.com) releasing the world’s top professional web site, Globally the most frequently visited website are Google and Yahoo, the most frequently visited Chinese website is Baidu. In 2007 report announced by market research company Metrics shows that Google has become the most frequently visited website by British and American smart phone users. From one aspect these data reflected searching their own interested information through using search engines occupied a very important position in their Internet lives, and have begun to infiltrate every aspect of people daily lives. However, with the popularization of Internet applications and increase in Internet users, information resources on the Internet have shown rapid index growth. Therefore, how to help users quickly and accurately find their own interested information of interest has become an important subject in the information retrieval study.This paper studied the user-oriented information retrieval systems theory and technology. User-oriented information retrieval research is raised mainly against the traditional systems-oriented information retrieval research which neglected user’s needs analysis. User-oriented information retrieval research hoped that through a series of measures such as research, mining, modeling of the users, needs, improved the design of information retrieval systems in order to further improve the efficiency of the user’s search. The full text is divided into six parts, including the following:The first part analyzed the cognitive processes of users needs in the course of information retrieval, including three stages:user’s needs cognition, information content cognition and information retrieval process cognition, and summarized the characteristics of user’s needs cognition processes; then discussed the deficiency of systems-oriented information retrieval research and information retrieval research model transformed from systems-oriented to user-oriented; Finally built a user-oriented information retrieval model, focused on the basic principles, the basic structure and key technologies of model construction.The second part focused on user-oriented information retrieval system how to mining users, information needs. Firstly analyzed the different characteristics of user’s needs in the long term and short-term, and discussed mining rationale of user’s needs according to the characteristics of user’s needs. Then introduced the common methods and process of user’s needs mining, and mainly analyzed the deficiency of the traditional needs mining methods. Finally proposed a client-based data collection improved program towards the traditional user’s needs mining method, and the technical implementation of the program. It’s necessary to emphasize that user’s needs mining improved method in this text didn’t completely replace the traditional methods. In fact the improved method existed deficiencies. This paper hoped that this method is a useful complement to the traditional user’s needs mining methods, explain problems from another angle.The third part major researched the user’s needs representation in the system. Firstly discussed the topic feature of user’s needs and the description of topic feature, and then introduced a common method of user’s needs representation, mainly discussed the user’s needs representation methods based on the topic feature. This part introduced the topic feature automatically extract technology, and according to the hierarchical relationship between topics, proposed the idea of topic hierarchical tree, and proposed a new feature evaluation function which learned from the idea of the classic TF-IDF weighting formula in the vector space model. Finally, this part proved feature extraction technique based on the topic hierarchical tree through experiments, and compared four kinds of feature extraction techniques, further verified the effectiveness of methods on user’s needs representation based on topic feature.The fourth part discussed the modeling of user’s needs. The modeling of user’s needs is in the core of the user mechanism in the user-oriented information retrieval system. In this part, the modeling process is divided into the initialization of user’s information needs, the studying of user’s information needs and the resolution of user’s information needs.This chapter firstly introduced the basic flow of the user’s needs initialization, focused on the cold start problem of user’s needs in the user-oriented information retrieval, and then put forward the method of centroid vector to solve the problem. Then introduced the application of relevant feedback technology in the user’s needs study, and used the classic Rocchio algorithm to update the user’s needs template. At the same time, improved the Rocchio algorithm according to the feature of the built user-oriented information retrieval model. Finally discussed the resolution of user’s needs. The purpose of this phase is to resolve the problem caused by a wide range of topic preferences of user’s needs, for example, not solution, will not be able to accurately grasp the topic feature of user’s needs. This part discussed the basic principles and algorithms description of the user needs to resolve the basic principles and algorithms described of the resolution of user’s needs.The fifth part described the implementation and evaluation of user-oriented information retrieval systems. Firstly, divided the entire retrieval system functional modules according to user management layer, information management layer and human-computer interaction layer, and simply introduced each module. And described the implementation details of some key features in the user-oriented information retrieval system; finally, evaluated the performance of retrieval system prototype this paper built. This part used two methods to evaluate performance. The first one was system-oriented information retrieval system performance evaluation methods based on the classical recall and precision ratio. In this method, in order to better adapt to the characteristics of the output in the retrieval system prototype, revised the recall and precision ratio. The second evaluation method designed an evaluation program which needed some volunteers to participate in according to interactive features of user-oriented information retrieval system. This evaluation method aimed at simulating a more realistic search environment, and reflected the retrieval performance change of the retrieval system in the interaction as far as possible.The sixth part is a case study, mainly introduced the applications of user-oriented information retrieval research in the agricultural information integrated service platform. Firstly introduced the target and system architecture of agricultural information integrated service platform. And then focused on the construct of user-oriented information retrieval system in the agricultural information integrated service platform, including the requirements of the retrieval system of agricultural information integrated service platform, the overall agriculture of agricultural information integrated service platform, the key technique analysis and the retrieval interface presentation of agricultural information integrated service platform. Specific analysis of this case provided some references for the practical applications of user-oriented information retrieval system.