Abstract: With the development of B2B E-commerce especially electronic marketplace, enterprises are provided many conveniences and flexibilities for electronic trading. But it brings the problem of information integration due to the heterogeneity of information, which becomes the bottleneck of B2B E-commerce development. The MAS B2B E-commerce model based ontology is introduced, which includes GMAg (Goods Matching Agent) to match goods. By hybrid considering semantic similarity of goods ontology’s part and attributes concept, a Structure semantic Similarity for goods matching is gives, which solves the heterogeneity of information for B2B E-commerce. At last, an instance of automobile proved this algorithm’s accuracy.

Key words: Ontology, Multi-Agents, B2B E-commerce, Goods Matching, Semantic Similarity

摘要： 随着B2B 电子商务特别是电子集市的发展,企业进行电子交易更加灵活、方便，但同时由于大量信息的异构性也带来了信息集成上的问题,而且这一问题已成为B2B 电子商务发展的瓶颈。文章建立一种基于本体的多Agent模型，增加GMAg(Goods Matching Agent)用于商品的识别和自动匹配，综合考虑商品组件概念和属性组件概念的语义相似度，提出基于商品本体结构语义相似度匹配算法，解决了异构商务系统之间商品信息自动匹配问题。通过一个汽车领域的实例计算，说明其匹配精度和效率。

关键词: 商品匹配, 语义相似度, 本体, 多智能代理, B2B电子商务