基于概念和统计的问答系统实现机制

问答系统与搜索引擎相比答案精练准确、“噪音”较少．为此给出了基于概念和统计的问答系统实现机制．首先确定焦点类型和答案类型，对于不同类型的焦点实行分级概念扩充；以段落为基本单元，利用潜在语义分析技术获取相关文本段落；给出了问题与句子的匹配算法和基于问题相似度的答案抽取算法，通过对语料进行潜在语义索引和增加FAQ反馈，不断增强系统的回答能力．依此研制的大学概况问答系统获得了较好的效果．

Implementation mechanism of question & answer system based on concepts and statistics

Question & answer system provides more accurate information with less noise compared with search engine.The implementation mechanism of question & answer system is presented based on concepts and statistics.It determinates the question focuses and the answer classes,and it makes a different conceptual expansion to a different question focus. Based on the text segment,it applies the latent semantic indexing to retrieve the related text segments.It gives the match algorithms between questions and the sentences,and it extracts the answers from FAQ using the similarity of sentences.It is able to update and improve the answering ability by LSI and FAQ.According to its mechanism of matching sentences and questions,the question & answer on introduction of Chinese universities is developed and proven to have a good precision.