中文阅读理解系统的研究与实现

阅读理解系统作为开发、评估和比较自然语言问答方法的可控测试平台,引起了自然语言领域越来越多学者的关注,设计并实现了一个中文的阅读理解系统,着重分析了问题分析、答案定位和答案提取等关键技术;并根据不同的问题类型,设计了不同的答案提取策略.实验表明,系统的性能比基准测试方法提高了近5个%.

Research and Realization of Chinese Reading Comprehension System

Reading comprehension systems have received increasing attentions within the NLP community as a controlled test-bed for developing, evaluating and comparing robust question answering methods. A chinese reading comprehension system is designed and realized, focused on the key technologies of the system, such as question analysis, answer location and answer extraction, and designed the different answer extraction strategy, according to different type of question. As the experiment illustrates, the performance of system improve approximately five percent than benchmark method.