Research On Technology Of Personalized Information Acquisition And Modeling

With the swift development and popularization of World-Wide Web, Internet has become the primary channel for spreading information, through which people can get various information conveniently. However, due to the huge amount of information and its explosive increasing rate, it is very difficult for people to locate what they need efficiently and accurately. Given such a mass quantity of resource, information retrieval is more and more important nowadays; nevertheless, the original retrieval system have a serious problem. It doesn't consider the knowledge background and the interest of the user, return the same result for the same key word of the retrieval user. The retrieval system should get the personal information speedy, for satisfying the individual information needs of the user whose needs are stable a couple of days, to supply personalized information service.Personalized information service systems widely search and acquire user information in both explicit and implicit ways, and construct user profile describing their interests according to this information. The more precise the interest description is, the more accurate information can be supplied to the user.This thesis first gives an in-depth discussion on the characteristics, architecture and three implementation methods of personalized information service system, then puts forward the integral design of the model, which consists of 4 functional modules-personalized information collection module, text-preprocess and feature-extraction module, user-interest-model-construction module and user-interest-model-update module. The latter chapters address the function and implementation of each module specifically, including key techniques involved, such as text auto-segmentation, part-of-speech tagging, feature extraction and machine learning, etc. Our focus will be mainly on three techniques, namely feature extraction, user-interest description, user-interest training and update.The characteristic of this model is its synchronized study with users based on user-machine interaction. Users will have access to their interest descriptions through interest training and update model, which they can modify directly...