Research On Virtual News Generation Technology Based On Text-Driven

Strategic Situation Representation (SSR) plays an important role in strategic decision simulation and is an essential part. Thus it becomes urgent to represent the changing situation during the decision simulation by the multimedia for constructing an"immersive"simulation environment in the field of strategic decision simulation. Traditional single situation representation can't satisfy the need of strategic simulation situation representation. In allusion to the problems such as low degree of automation for strategic situation representation, the concept of virtual news is introduced in this thesis.Virtual News is a novel representation form in SSR. Narrative representation technology is applied to transform the textual description of strategic situation into virtual news that is revealed to the decision-maker. Virtual news should be capable of quick generation, which is determined by the characteristic of SSR. The primary research of this thesis is automatic parsing of textual content and video search ranking technology.To the question that the characteristic of news, the conception of pictured entities is presented in this thesis, and a narrative template to normalize textual content is established, and several methods to modify template are presented. Further more, in allusion to the truth that it's difficult to support retrieval examples, news material ranking based on mutual reinforcement principle is presented.A generation system of virtual news is designed on the foundation of analyzing and studying the technology. The system has the function of automatic parsing of textual content, generating normalized narrative template, searching the most proper news material, and then generating integrated virtual news with proper representation format. The performances of the system show that algorithms presented in this thesis are reasonable and present the characteristics of the strategic situation well.