Research On Topic Tracking And Evolution Analysis Technique

Topic Tracking and Evolution Analysis is an event-based information organizing task for tracking the reappearance of the news topic user interested and analyze the evolution process of the topic. Its purpose is to organize information efficiently and help people finding what they want intuitively. It is theoretically and practically valuable in military and civil field. This dissertation studies the topic tracking algorithms, topic evolution analysis method and event discovering and evolution analysis technique. The major contribution of this dissertation is as follows:(1)After the analysis of topic's structure together with the concepts related, we propose a Topic Structure Model constructed by news layer, subtopic layer and event layer. Considering topic through such a structure view point instead of the traditional flat news layer could take more issues into account. The setup of the model lays a foundation of the topic tracking and evolution analysis.(2)The causation of the topic evolution is the shifting of the various aspects in topic. Based on this conclusion, this dissertation proposed a Subtopic Feedback based Topic Tracking Algorithm, which utilize the idea of New Event Detection to organize the news temporally. The experiment in this dissertation proves the algorithm could adapt the evolution of the topic commendably.(3)After the analysis of the topic evolution, this dissertation proposed a Subtopic Similarity based Topic Evolution Analysis Algorithm. This algorithm analyze the evolution of the topic though the relation of the subtopic, and could show the process of the topic evolution in the granularity of time slice. The experiment in this thesis proves the effectivity of this algorithm.(4)Based on the Topic Structure Model, this dissertation proposed an Event Discovering and Evolution Analysis Algorithm. The algorithm takes the time, location, character and organization properties into account. The experiment in this thesis shows the algorithm is effective.Finally, the design principle and technique of realizing cartographic generalization system are brought forward, the research work of this paper is summarized, and the future developing direction of the Topic Tracking and Evolution Analysis technologies is indicated.