Business process global conformance checking method based on BERT
This paper proposed a BERT-based conformance checking method to assess the fitness and precision of process models.The method used the BERT model to generate vector representations of control flow variants,the evaluated conformance by comparing the similarity between traces from event logs and replayed model logs.The introduction of the BERT model enabled the capture of fine-grained behavioral features,addressing the limitation of traditional shallow representation methods that primarily extract static behavioral information.Experiments conducted on real event logs demonstrated that this method performed well in fitness and precision evaluation,providing a more detailed reflection of the alignment between process models and observed behaviors.This showed that the method had significant advantages in uncovering dynamic behavioral features,offering a more accurate evaluation tool for conformance checking in process mining.
万方数据
维普
