用自然语言处理改进问答系统的研究和实现

问答系统是当前人工智能应用的一个重要领域，而传统的基于模式匹配方法的问答系统具有很大的局限性。该文研究了基于HMM模型、图句法分析、词典和规则的自然语言处理算法。将这些算法应用到问答系统中，扩展了对话管理模块，对用户的自然语言进行语义分析，从而实现对话的语义块识别、主题识别和对话信息提取，提高了系统对复杂用户输入的处理能力，克服了传统方法的不足，并使用Java实现了一个实验系统。

Study and Implementation of Improving QA Systems Using NLP

Question answering is an important field in artificial intelligence researches. But there are some limitations in traditional QA systems based on pattern matching. This paper analyzes and applies natural language process algorithms based on HMM, chart parsing, word dictionary and syntactic rules to extend dialogue management module, perform semantic analysis on users’ sentences to implement semantic blocks recognition, theme recognition and information distillation. Those algorithms can improve system’s process ability towards sentence analysis and overcome the disadvantages of traditional methods. The QA system is implemented on Java platform.