Towards Activity Representations for Describing Task Dynamics

摘要

Activity representations are proposed as an extension to traditional task models. Basically, an activity representation describes fragments of knowledge about several tasks and how to interleave or merge them. Knowledge about single tasks is spread over several representations at different levels of abstraction. Lower-level models are more ephemeral and help people to organise their day-to-day activities. On the one hand, their creation is supported by more stable representations reflecting goals, activity rhythms, domain knowledge etc. On the other hand, situated action is necessary to create such (task) knowledge. We show that higher-order activity representations provide a better explanation of some task-related aspects than monolithic single task models. For example, they support re-/on-the-fly planning and contribute to dispel the belief in complete and consistent task descriptions. The paper focuses on task redefinition, task grouping and polymotivated actions, activity spaces, goal elaboration, and the interplay between habits and learning. Some conclusions for interaction design are given.