Multiple Text Summarization Technology-based Web Enterprise Competitive Intelligence Service Platform

Intelligence briefings can reflect the value of information in time, support the managers to make routine decision, establish the communication bridge between the information department and the manager. Automatic generation of intelligence briefing can overcome the disadvantage of low efficiency of the manual way. It both reduces the staff's workload, and improves the service quality of intelligence. How to generate fast and accurate summary of intelligence, and improve the efficiency and quality of information retrieval services effectively, thus allow users easily access to valuable intelligence information, is the main content of this article studying.Based on the project "Competitive Intelligence Service Platform and Service System Construction in Yun Nan province" and "Competitive Intelligence Management System in Business", this paper makes a brief description of the related technology, gives a brief analysis of the current development of the research and application of CISP both domestic and abroad, conducts a detail analysis of business requirements and in-depth study of automatic summarization techniques based on subtopic. Based on this, this paper designs the automatically generate strategies of intelligence briefing and implements it in the project.This thesis studies the key technologies of automatic summarization techniques based on subtopic, such as subtopics divided, subtopic sorting, sentence extraction and sentence sorting. By employing the maximum and minimum distance clustering methods to determine the amount of subtopics, the results of divided subtopics would reflect the true content of the text more accurately. Such designing of abstract summarization scheme combining subtopic ordering and dynamic sentence weights adjustment could reveal the text contents more comprehensively. This thesis gives a sentence sorting method, which combines the text frames and the importance of intelligence order, basically solves the multi-document summary sentences sorting problem. Multi-document automatic summarization could be realized and provide practical foundation for system intelligence briefing automatic generation.Combining the multi-document summarization based on subtopic segments and the intelligence topic structure, this paper designs and implements the strategy of intelligence briefings generating. Because of many kinds of intelligence briefing automatic generating, it improves the intelligence analysis and intelligence service efficiency which provides support for senior management decision-making.