Purpose/Significance To construct a data mining system integrating domain prior knowledge，and to enhance the discov-ery efficiency of core compatibility rules in traditional Chinese medicine(TCM)prescriptions. Method/Process Based on the W-WD-GetRules algorithm，“Jun-Chen-Zuo-Shi”theory in TCM is integrated，an item weight mechanism is constructed，and a weighted downward closure property(WDCP)pruning strategy is introduced to perform weighted mining on high-dimensional sparse prescriptiondata. An analysis platform integrating dynamic weight configuration module and an asynchronous mining engine is constructed based onthe B/S architecture，utilizing force-directed graph visualization to present compatibility patterns. Result/Conclusion The system can ef-fectively reduce the interference of high-frequency adjuvant herbs and reflect the clinical logic of TCM prescription formulation，provid-ing a new path for solving the“homogenization”challenge in TCM prescription data mining with both mathematical rigor and medical in-terpretability.