With the fast growth of B2B sales, an intelligent system is greatly useful for decreasing transaction cost and increasing market efficiency on electronic platforms. In order to improve the quality of transaction processing and customer experience, this paper proposes a knowledge-based system, which employs a Case-Based Reasoning (CBR) technique for trade matching in B2B marketplace as a substitute for the manual matching process. The system function and logical architecture are discussed. And the case repository is proposed to support this CBR approach where the case representation, case base indexing, case base decomposition and the dictionary are argued in details.