【摘要】 基于自动构建的问句模板库,对中文复杂类问题进行问题分类和关键词抽取,并提出一种基于在线百科知识和搜索引擎的关键词扩展方法,从而提高了系统的整体表现。实验采用NTC IR-7 CCLQA的测试语料,表明该方法的问题分类准确率达88%。关键词的扩展使得系统F值提升了20%。更多还原

【Abstract】 To deal with the Chinese complex questions,this paper introduces a question classification method and extraction based on the question pattern library which is built automatically,and a keywords expansion method is proposed based on online encyclopaedia and searching engine in order to improve the QA system performance.The test corpus of CCLQA in NTCIR-7is used in experiment,and the result shows that the question classification precision is up to 88%,while F-value of QA system improved 20% in case of keywords expansion.更多还原