招生领域问答系统中问题理解的研究硕士学位论文.doc

摘要 PAGE I 摘 要 随着互联网的高速发展，网上的信息越来越多，如何在这些海量信息中快速准确地找到所需要的信息也越来越困难。自动问答系统的出现很好的解决了这一问题，它既能够让用户用自然语言句子提问，又能够为用户返回一个简洁、准确的答案。中国每年高考的考生数百万之巨，很多考生和家长对高考有这样或那样的问题。因此，开发一个招生咨询领域的自动问答系统是非常具有现实意义的。 本文通过对自动问答系统中的答案抽取部分的研究，提出了一种将FAQ库（常问问题答案库）与蕴涵丰富招生咨询材料的文本库相结合的答案抽取方式，有效地提高了系统的性能。在FAQ库中，由于存储了大量而正确的问题答案对，使得答案的抽取更为准确和快捷；在基于FAQ库的答案抽取方法中，本文提出采用基于关键词信息和基于语义词典相结合的融合算法来计算用户问句与FAQ库中问句的相似度。该算法不仅考虑了词面的信息，更深入到语义层次的信息，使句子间的相似度计算更为精确，从而使答案的提取也更为准确。在文本库的答案抽取方法中，采用的是比较成熟和常见的基于向量空间模型的TF-IDF方法来计算问句与文本之间的相似度。并将从文本库中提取出的答案定期进行人工判断整理，备份至FAQ库中，从而使更多的问题能直接从FAQ库中获得答案，有利于提高系统的运行速度。 最后本文设计实现了一个招生咨询问答系统。实验结果表明，系统可以快速、准确的满足用户的需求，从而有效地验证了上述方法的有效性和可行性。 关键词：自动问答系统；答案抽取；常问问题答案库（FAQ库）；文本库 ABSTRACT ABSTRACT With the rapid development of Internet, there are more and more informations online, how to find the informations they need quickly and accurately are more and more difficult.The Automatic Question and Answering System is a good method to resolve this problem,it not only let people ask a question with natural language,but also return a concise and accurate answer.China has millions of candidates take the university entrance examination each year, many students and parents have this or that problems on the college entrance examination.Therefore the development of Question Answering System based on Admission Counseling has a very practical significance. In this paper, research by Answer Extraction of Automatic Question and Answering System,it advances a new method that combines the Frequently Answer Question Library (FAQ) with the document library which has the rich materials about Admission Counseling,and effectively improves the performance of the system.In the FAQ Library, it can get the answer accurately and quickly because it stores a lot of correct question-answer pairs;In the method of Answer Extraction based on the FAQ Library, the paper advances the amalgamation arithmetic that combines the keywords information with the semantic dictionary to calculate the similarity between the user’s question an