Research And Implementation Of Scientific Social Network Based On User Interest

The rapid and effective development of Inforamtion Technology and the Internet produces great convenience for scientific researchers, but also accompany with so called"information overload"and"information lost"problems. The rapid arising of usable literatures in scientific areas makes researchers more difficult to find the literatures they need for their research, on the other hand, researchers always do repetitive searching in order to get literatures due to the lack of effective sharing and communication mechanisms, and these repetitive searching lower their research efficiency.Currently, researchers always use literature management software to organize and manage literatures, so every single researcher has a personal library. If we unite all the personal libraries together, the big global library will be a good base for sharing literatures on one hand; on the other hand, we can implement personalized literature recommendation with user interest model which are obtained by processing and digging literatures in the personal literature library of every researcher. This thesis penetrates analysises the searching and sharing methods of researchers and the communicating channels between researchers; puts forward a Scientific Social Network (SSN) which is constituted of scientific researchers and based on bottom peer-to-peer network; define the concept, architecture, behaviors and characteristic parameters of SSN.User interest model must be established before build up the SSN. Considering from the realistic requirement of SSN and the special characters of the researcher colony, the thesis investigates and improves on the actuality of user interest modeling and personalized recommendation technology at first. Based on the study, a user interest modeling algorithm based on vector space model (VSM) has been proposed, built up the interest model of researcher and interest tendency model of literature, an experiment has been done to validate the effectiveness of the new user interest modeling algorithm, the thesis presents the input data, steps,results and analysis of the experiment, it shows that VSM based user interest modeling algorithm is effective basically, but still exists disadvantages such as lack of stability.SSN is some kind of a complex network, based on the user interest model, we study the SSN which consider researchers as its nodes and the relationships between researchers as it edges deeply from the complex network view, the relationships of researchers are determined by user interest model; then a topology build algorithm has been proposed and the emulation experiment has been done to validate the effectiveness of the algorithm. The algorithm contains three procedures which are static network generation, adding new peers and network evolution after adding new peers. The background, data set, methods, setps and results of the experiment have been presented, then we present the degree distribution and average path lengh of the ssn, the analysis of the experiment result shows that the SSN is a small-world network. The problems in inefficient searching in researchers'colony have been solved properly.Finally, based on the four key technologies researched in this thesis, the requirement of SSN has been analyzed, the architecture and modules of SSN have been designed and the prototype of SSN which is overlayed on the bottom peer-to-peer network has been implemented.