面向中文工程论坛问答的知识应用情境获取及应用.pdfVIP

计算机集成制造系统 第 22 卷第 5 期 Vo l. 22 No. 5 201 6 年 5 月 Computer Integrated Manufacturing Systems May 2 0 1 6 DOI :10. 13196/j. cims. 2016. 05. 003 面向中文工程论坛问答的知识应用情境获取及应用 李心雨，蒋祖华+，宋 博，王海丽，孙小明 (上海交通大学机械与动力工程学院，上海 200240) 摘 要:为了高效、快速、准确地处理工程论坛问答中未经规范描述的知识应用情境，面向工程论坛中的中文 问答记录，提出一种人工干涉少、用户需求敏感性高、情境匹配准确性高的知识应用情境获取方法，进而从自然语 言描述的文本中快速获取与用户当前任务关联的知识应用情境并推送相应答案。以 CAD 技术论坛中的问答为实 验素材，对工程师可能产生的知识需求做出预测。实验表明，该方法比现有方法能够更有效地利用中文工程论坛 问答这种普遍存在但开发不足的知识资源，快速准确地获取其中匹配用户知识需求的知识应用情境和答案，有助 于低成本地建立一套主动式知识推荐系统并提高其工作性能。 关键词:知识应用情境;论坛问答;文本挖掘;知识推荐 文献标识码:A 中图分类号 :TP391 Retrieving and applying knowledge application contexts from forums questions and answers in Chinese engineering LI Xinyu , JIANGZuhua+ , SONGBo , WANG Haili , SUN Xiaoming (School of Mechanical Engineering , Shanghai ]iaotong University , Shanghai 200240 , China) A恼tract: To conduct the un-annotated informal Knowledge Application Context (KAC) in forum Question and An swer (QA) efficiently , rapidlyand precisely , a KAC elicitation method which had higher sensitivity and precision in matching KAC with less human intervention was proposed from natural-Ianguag e-describing forum QA texts and recommend corresponding answers related to users current missions. KAC retrieving experiment was carried out in CAD technical forum , and thè correct prediction of engineers knowledge needs was conducted. The result in dicated that the proposed method significantly outperformed than the currents methods in utilizing the QAs in Chi nese engineering f