Abstract: Combining distinct sources of evidential knowledge in support of a ranking can improve the retrieval performance effectively.Adopting synonym-based evidence to extend the basic belief network can obtain an extended belief network retrieval model.The extended model uses different strategies to combine evidences leads to different ranking computation formulas，then influences the retrieval performance.This paper gives conjunctive combining algorithm based on disjunctive combining algorithm，and some proper modifications are given as to the conjunctive combining algorithm.Experimental results indicate that the retrieval performance based on conjunctive algorithmis higher than that based on disjunctive algorithm.

摘要： 依据《同义词词林（扩展版）》，以初始查询的同义词为证据扩展基本信念网络，得到扩展的信念网络检索模型，扩展模型采用不同的归并方法将得到不同的排序计算式。在析取归并算法的基础上，给出了合取归并算法，并对其进行了合理的修正。实验证明，合取情况下的检索性能高于析取情况。