In this thesis, various data mining methods are integrated to construct the on-line rescheduling system and the decision support system for manufacturing environment. For on-line rescheduling system, an interval variant rescheduling mechanism is proposed. In order to deploy different dispatching rules to different intrabays, k-means is used for clustering the intrabays of the fab. Then genetic algorithm (GA) is employed for searching dispatching rule sets which promote better performance. In terms of the system conditions corresponding to dispatching rules, features can be extracted through generalized discriminant analysis (GDA) and two kinds of classifiers, KNN (K-Nearest Neighbors) classifier and SVM (Support Vector Machine) classifier, are constructed as schedulers. In addition, the ANFIS (Adaptive Neuro-Fuzzy Inference System) prediction model is built for the sake of on-line deciding the scheduling intervals. The experiment results indicate that applying the proposed mechanism to obtaining dispatching strategies is an effective method considering the complexity and variation of semiconductor wafer fabrication systems. The decision support system communicates with the two fab models (SEMATECH model and TRC model) and contains three subsystems. They are rush order handling subsystem, diagnosis and maintenance subsystem, and knowledge management subsystem. In particular, the methods for knowledge extraction, learning, and update are provided. Also, four scenarios are provided to support decision making. The first offers decision makers to decide product mix ratio with the concept of TOC (Theory of Constraints). The second can control job arrival rate through the monitoring of WIP (Work In Progress). And then the third one decides dispatching rules in terms of the knowledge. The last scenario aids preventive maintenance with the information of each machine’s PM schedule. All these scenarios are through web pages to achieve knowledge sharing.