Exploiting neighborhood and multidimension granular information for supporting design rationale retrieval

Based on our previously proposed ISAL model (issue, solution and artifact layer) for design rationale (DR) representation, in this paper, we report our efforts in researching an ISAL based DR retrieval framework to better support DR retrieval by taking advantage of neighborhood and multi-dimension granular information presented in DRs. In our proposal, DR is firstly extracted and indexed using ISAL and a document profile model respectively. Next, an initial DR graph is formed by linking up different DRs based on their document citations and document similarities. A DR network is therefore established by integrating similarities from issues, solutions and artifact aspects using neighborhood information in the DR graph. In order to prioritize DRs retrieved, a graph-based ranking approach is further engaged. To validate the approach proposed, we have reported our preliminary experiments on issues like DR indexing based on different approaches, similarity measurement in DR network, and lastly, a brief example of using neighborhood information to suggest potential DR related concepts in retrieval query processing.