Take a Break in the Middle: Investigating Subgoals towards Hierarchical Script Generation
This work addresses the problem of generating structured, goal-oriented scripts for AI systems, but it is incremental as it builds on existing tasks with new hierarchical benchmarks.
The paper tackles hierarchical script generation by proposing a dataset and baselines that incorporate subgoals, showing effectiveness through automatic and human evaluation, though subgoal discovery is found to be more challenging than summarization.
Goal-oriented Script Generation is a new task of generating a list of steps that can fulfill the given goal. In this paper, we propose to extend the task from the perspective of cognitive theory. Instead of a simple flat structure, the steps are typically organized hierarchically - Human often decompose a complex task into subgoals, where each subgoal can be further decomposed into steps. To establish the benchmark, we contribute a new dataset, propose several baseline methods, and set up evaluation metrics. Both automatic and human evaluation verify the high-quality of dataset, as well as the effectiveness of incorporating subgoals into hierarchical script generation. Furthermore, We also design and evaluate the model to discover subgoal, and find that it is a bit more difficult to decompose the goals than summarizing from segmented steps.