Towards Automated Synthesis Of Executable Eclipse Tutorials

Eclipse tutorials guide users step by step to perform programming tasks. However, as current executable tutorials can only guide users through one way, users' specific requirements cannot be satisfied. In this paper, we propose an automated approach to synthesizing different tutorials to generate a tutorial that covers many ways. We use record and replay techniques to generate the original tutorials and synthesize them by mining experts' comments and actions. First, we find the same sub-tasks of different tutorials according to their titles and synthesize the actions of a sub-task by building a hidden Markov model. Then with the synthesized tutorial, steps of a task are recommended gradually during the programming process and executed automatically by replaying some actions. In addition to presenting this approach, we finally present a tool to implement the approach and evaluate its feasibility by an experiment.