Understanding task structure in DSM: Mining dependency using process event logs

Abstract

Dependency Structure Matrix (DSM) has been widely used to analyze and present the inter-structure of design projects which are often characterized by many interrelated tasks. One essential role of DSM is to reveal the dependencies amongst different tasks. To succeed, it heavily relies on how well the initial dependencies are identified. Conventionally, it is accomplished by engineers through interview, survey and discussion that are obviously constrained by resource available in performing such activities. Hence, its liability heavily relies on the understanding of engineers. Moreover, for a complex design project where a higher number of tasks are involved, to gain a comprehensive understanding of their intricate relationships is also a non-trivial task. In order to tackle this, in this paper, we propose a flexible approach to mine dependency from process event logs which dynamically record the detailed information of task execution in a real context. Using process event logs, a number of dependencies can be steadily discovered and derived by focusing on different subsets of data for specific purposes. In the end, a case study is used to illustrate the proposed approach.