Abstract: For the knowledge management in runtime product development business processes,a context-aware workflow knowledge recommending approach was proposed.Based on the investigation of workflow task execution contexts,a hierarchical workflow context ontology model named WfCOM was proposed.Aiming at the common features of elemental and structural levels of WfCOM instances,three algorithms were respectively studied:a linguistic similarity calculation algorithm for string similarity measuring based on Jaro-Winkler,a text vector space similarity calculation approach based on TF-IDF for ontology description,and a structural self-similarity calculation method for context ontologies based on graph theory.Based on the presented ontology matching approaches,a linear combination approach to calculate the multi-dimensional context similarities was proposed,and a weighting vector searching approach was also proposed based on access history of business artifacts,which was then evaluated thereafter by experiments.

Key words: context ontology, natural language processing, business artifact, product development process, business process, knowledge management

摘要： 针对产品开发业务过程任务执行时的知识管理问题,提出了一种情境感知的工作流任务知识智能推荐方法。在对工作流任务的执行情境进行研究的基础上,建立了一种基于本体论的分层工作流情境本体模型WfCOM,并对其进行了形式化描述;针对WfCOM情境本体实例节点在元素层面和结构层面的共有特性,分别研究了基于Jaro-Winkler的语言学字符串相似度算法,基于TF-IDF的本体描述文本向量空间相似度算法,以及基于图结构自相似的情境本体结构相似度计算方法。在上述本体节点匹配算法的基础上,研究了一种线性组合的多维度情境相似计算方法,为了找到最优权重向量,研究了一种基于业务产物访问历史评价的权重向量搜索方法,并通过实验对提出的多维情境相似算法进行了验证。

关键词: 情境本体, 自然语言处理, 业务产物, 产品开发过程, 业务过程, 知识管理