The Research Of The Active Infromation Retrieval Method Based On Group Collaborative Discussion

Group collaborative discussion is a effective solution to solve the complex problems in the information society and is also one of the important tools in collaborative learning in the field of education. The revolutionary changes have taken place on the models, the methods and the technical means of group collaborative argumentation with the rapid development of modern science and technology.This paper has researched designing and implementation of the tools supporting on-line group collaborative discussion. The main work of this dissertation focus on designing and implementing the tools supporting on-line group collaborative discussion based on the methodology of meta-synthetic engineering, which combines the active information retrieval method, the information technology and the intelligent technology together. The group collaborative discussion system contains three subsystems: the opinions expression and induction sub-system, the topic detection, tracking and collaborative filtering recommendation sub-system and the scheme creation and evaluation sub-system. The opinions expression and induction sub-system supports the whole argumentation process and provides the data sources for the topic detection and tracking and collaborative filtering recommendation sub-system. The topic detection and tracking and collaborative filtering recommendation sub-system obtains the argumentation information from the opinions expression and induction subsystem and analyzes them, which provides web information resources for the groups in the argumentation process. The scheme creation and evaluation subsystem provides optimum scheme in the end stage of discussion. The three sub-systems are indivisible and complement each other. The main contributions of this paper include:1. First of all, this paper generalizes the overview and the definition of on-line group collaborative discussion, and then summarizes a number of the on-line group collaborative argumentation systems and analyzes their disadvantages. It also elaborates the research overview of active information retrieval.2. A new opinions expression and induction sub-system to support the group discussion is designed and implemented in this paper, which provides an effective tool of opinions expression and online discussion through computer and network. This paper adopts the structure speech model, firstly, a dynamic speech tree is built during discussing, which reflects the logical relationships among speeches and makes it easy for computers to deal with the speeches,and then opinions are automatically analyzed and gathered by the system, and opinions are merged according to the similarity.3. A framework of topic detection, tracking and collaborative filtering recommendation sub-system is designed. First of all, the hierarchical dictionary is established based on the knowledge of experts in the field, which is the key of the topic detection and tracking. Secondly, the word frequency of the word segmentation results in the opinions expression and induction subsystem is calculated, and then topic vectors are obtained based on the field dictionary. The topic tracking is implemented by calculating the similarity of the two topic vectors. Finally, the websites with a greater correlation are automatically recommended to the users who need them by the recommended model.4. A scheme creation and evaluation sub-system is proposed in this paper. The schemes of experts are automatically analyzed and gathered through electronic questionnaire distribution and collection tools of this system. An efficient method to evaluate the schemes is also proposed, in which the qualitative evaluation results can be quantized by the consensus algorithm. Then, the evaluation results are conveniently visualized by the Isomap algorithm. Finally, A optimal scheme is obtained during the repeated discussion.5. This paper discusses the overall framework and the operating environment of the active information retrieval system based on the C/S structure, and the key technologies are also discussed, such as .NET, XML and the visualization technology based on ZedGraph.