BESMER: An Approach for Bad Smells Summarization in Systems Models
Zhao Xin, Jeff Gray · 2019
Bad smells are surface indications of potential problems with source code quality and have been investigated deeply within the purview of object-oriented programming. However, there has not been much research conducted to understand bad smells within the context of systems models. Moreover, the majority of bad smells in existing literature have been suggested by experienced developers and researchers who may view "smells" differently from inexperienced developers. To this end, we propose BESMER, our project that categorizes bad smells in systems models by mining discussion forum posts of end-users of a specific systems modeling tool. Specifically, this paper describes how our three-level discovery mechanism based on machine learning techniques assisted us in finding bad smells from the LabVIEW online discussion forum. Our experimental results not only confirm that end-users also encounter the bad smells proposed by experts, but also reveal new bad smells for LabVIEW models. We also present some implications discovered from the examination of user posts and list areas of future work based on our current findings. As far as we know, this is the first paper to investigate bad smells from an end-user's perspective within the context of systems models.