Challenges and Solutions for Hierarchical Task Network Planning in E-Learning
Carsten Ullrich, Okhtay Ilghami · Starting AI Researchers' Symposium · 2006
This paper describes a collaboration between two PhD students. The first author is developing a course generator (CG) for a Web-based e-learning environments (WBLE) [3]. A CG generates sequences of learning objects according to the learner’s goals and individual properties. The assembling process is knowledge intensive: assembling courses that implement modern pedagogical theories requires a framework that allows to represent complex pedagogical strategies. The strategies are implemented using the Hierarchical Task Network (HTN) planner JSHOP2 [1], developed by the second author. In HTN planning, the goal of the planner is to achieve an ordered list of top tasks, where each task is a symbolic representation of an activity to be performed. The planner formulates a plan by applying methods that decompose these top tasks into smaller and smaller subtasks until primitive tasks are reached that can be carried out directly. When we applied the HTN framework to WBLE, we encountered several challenges, which we will describe in this paper, together with our solutions. We believe that this work is of general interest, as today’s WBLE are excellent examples of distributed information systems and illustrate the challenges that can arise in complex scenarios.