Process Mining Method Based On Log Abstract Research

The development and application of modern information technology pays much more attention to the specification and transformation of business processes than ever. Process mining is the extraction from the process log process information in order to build a clear process model, which ensures that the process model is consistent with the execution of actual process, to describe and analyze the existing business processes to support process redesign.The traditional process mining algorithms cannot handle unstructured business processes, and the model constructed by existing algorithm is not conducive to analysis effectively. In this thesis, we adopt the abstraction of pattern to preprocess the logs, improving fuzzy mining algorithm to construct a hierarchical process model. This method provides a reference for the analysis and reconstruction of business processes, made better use by process management systems and users, and improve the work efficiency of business processes.In the pre-processing stage, a pattern-based process log preprocessing algorithm has been proposed. According to the defined mode selection indicators, the log has been classified and clustered, making the log as much as possible with less noise. Using the abstract model to represent the log, the structure of the model can be simpler. The improved fuzzy mining algorithm defines of the basic indicators, proposes a two-phase approach to deal with the abstracted logs, then draws the model’s node graph, and clusters the aggregation model. Thus the global process model has been obtained, which can also be enlarged local sub-processes, providing a basis for further process analysis. Finally, based on the ProM platform, this paper develops two plugs, one is based on the abstraction algorithm, and the other is based on the improved fuzzy mining algorithm. The correctness and validity of the proposed algorithm is verified by experiments.The study results show that the proposed algorithm can eliminate the irrelevant information in the log during building the process models, and reduce the complexity of the process, improve the understandability of the model. Therefore, the algorithm in dealing with complex and unstructured process mining has obvious advantages.