摘要
交通知识与人的生命安全息息相关。针对如何方便快捷的获取交通知识,设计并实现了以即时通讯软件微信为人机交互媒介的移动智能自动问答系统。首先,对文本进行特征向量提取,并对同义词进行归一化,消除同义词对查询准确率的干扰;然后,综合词频和词性信息计算文本关键特征的权值;最后采用BM25模型计算问题与知识库中文本信息的相似度,返回与问题最相似的答案。实验表明,本系统的移动性强,人机交互友好,查询准确度高。
Traffic knowledge is closely related to people's life and safety.Aiming at how to obtain the traffic knowledge conveniently and quickly,the mobile intelligent question answering system,which is based on the instant communication software We Chat,was designed and implemented. Firstly,extract the feature vector of the text,and normalized synonyms,eliminating the interference of synonyms on query accuracy.Secondly,calculation of critical feature weights according to the part of speech and word frequency.Finally,according to the BM25 model to calculate the similarity of text,return the most similar answer to the question. Experiments show that the system features strong mobility,man-machine friendly interaction and high query accuracy.
出处
《电子测试》
2016年第6X期25-28,共4页
Electronic Test
关键词
智能问答
微信
BM25
同义词归一化
文本相似度
intelligent question and answer
We Chat
BM25 model
normalized synonyms
text similarity
