基于全信息理论的短文本倾向识别研究

Based on the theory of Comprehensive Information Methodology - Natural Language Understanding (short for CIM-NLU), a short text orientation identification system for multi-domains was realized via Multi-agents. Modules able to learn the orientation of short text were implemented respectively for assigned domains, using the comprehensive information knowledge-bases made up of syntactic, semantic and pragmatic information. The result of the experimental system proves the feasibility, effectiveness and great potential of this method.