Analysis and classification of task knowledge patterns

Patterns have recorded the experience of engineering software systems. Various patterns have been introduced and described in several domains. A particular class of patterns, task knowledge patterns that are important for agent oriented software development, is investigated in this paper. It has been reported that descriptions of task knowledge given in the pattern literature is not clearly structured, and some of the useful description is left implicit to the developer. This paper reports our investigation on task knowledge that has been shared across a wider spectrum of articles to showcase such scenario. We present a new way to analyse and classify task knowledge patterns. Furthermore, we demonstrate how the experience of different development groups is described very differently based on the classification result. The investigation shows the need to present task descriptions in a structured manner. The analysis and classification techniques demonstrated are applicable for agent-based simulation