Research Of Flow-optimized Technique Applied In Workflow

Nowadays, a major limitation of current workflow systems is their lack of the support for dynamic workflow adaptations. However, this functionality is a major requirement for next-generation systems in order to provide sufficient flexibility to deal with unexpected situations and failures, and it is a hot issue which is studied currently. In this paper, the issue is studied and several ideas are put forward to support the dynamic workflow adaptations.As aspect of supporting for exceptions or changing, this paper pointed out the defects of current Workflow Systems, and analyzed the results of research of supporting dynamic adaptability. Aiming at the special background of dynamic changes, such as the workflow management in the web-based teaching, in which the work flow is dynamic, and need to be changed according to the users and situations, enhancement was made in supporting for dynamic adaptability by improving the case-based way of learning and optimizing the work flow.According to the idea of the 'Activity Theory' and 'machine learning', the model of supporting for dynamic adaptability in the workflow was put forward. The case-based way of learning was applied to induce the structure of the workflow model. In the course of inducing, the structure of the workflow model is showed with HMM (hidden markov models). Furthermore, the corresponding operators of merging and splitting are defined for acquiring the appropriate HMM. So the work flow was optimized by merging and splitting. The results of inducing were used for intelligent prompt with a list, which would be updated with the workflow instance being accomplished.According to these ideas, a workflow management system oriented the web-based teaching was implemented, which had the characters as following: Having the primary ability of learning. Supporting for the Ad Hoc work flow. Running without models at first. Prompting with the intelligent list, so that the times of interaction were decreased.