Learning a and predicting workflow for a certain goal

I am trying to come up with a strategy which could learn a workflow for various goals and then based on some input goal, predict the workflow which could fulfil the goal. Goals are achieved by workflows. E.g., Reach Office:= Tasks(‘Leave Home’, ‘Travel to Location A’, ‘Board Bus’, ‘Drop at Location B’, ‘Walk to Location C’). I wish to understand how a learning mechanism would enable learning such multiple goals and their corresponding workflows. And once the system has learnt the goals and workflows, the system should predict workflows for new goals like "Reach Theater", "Make Coffee" etc. The training data will include lots of existing goal and task recipes. Any pointers in this area would be very helpful. Can anyone suggest any existing approach?