Research Of Personalized Independent Process Construction Method

Despite of the gradual development of workflow technology from the initial sprout, and achieved considerable success, from the practical applications status of workflow perspective it is still far from the level one would expect. Because of the presence of a large number of the interim decision and unstructured companies business processes, in the traditional way of model defines in advance, it is a very difficult thing that if you want to make timely changes to the activities which are in a submitted workflow instance. This situation will greatly limit the application of the workflow management system in the enterprise. However, currently in most of the workflow management systems, improving the flexibility of the process often declines the support of application system. So the problem how to improve the flexibility of the workflow management system effectively gets attention of the industry and researchers.For improving flexibility and contradiction of process flexibility and support in workflow management system, paper presents personalized independent dynamical process construction method based on users’ features and preferences. The method extends the service-oriented collaborative filtering to more complex similarity model for process execution instance. In every implementation step of the process, this model selects or recommends activities for the current user to achieve the entire process dynamically based on similar users’ track and execution context of process instance. This method does not require pre-modeling and improves the constraint-based workflow management system. It achieves a balance between flexibility and support in workflow management system. Also it improves the overall process quality and user experience.Firstly, introduce personalized independent process construction model. In independent process construction, this model is based on user goals, domain knowledge, process execution history logs, and user features. User goals is the task required to complete; domain knowledge is used to establish abstract activities which are used classify activity and build constrain the model; process execution history includes the log of uses’track and log all users’ feature; user features refers to the current user’s part trace and the log. With the above information, constraint model engine calculates the active set recommended candidate; similarity calculation engine computes the users’ similarity based on the active set; activity evaluation and recommendation engine calculates recommended degree of process tracks.Secondly, introduce personalized independent process construction method based on process logs in detail. The method is based on log analysis. First, this paper presents two mechanisms used extract process logs, prefix extraction and set extraction.The model calculates matching between log tracks and part tracks for filtering workflow logs. Then calculate the similarity of user preferences based on these misplaced workflow logs. For recommended activities set, if the task in user’s goal is same with the before, the recommended activities set is calculate by abstract activity; else by workflow logs. Finally, considering the uses’ similarity and users’ preference value of the candidate activity, the model recommends optimal activity tracks.Finally, describe personalized independent process construction method and experimental analysis.