Discovering traceability between business process and software component using Latent Dirichlet Allocation

Software system is built to support business process. Software system needs to evolve over time because there are some changes on business processes. A relationship exists between business processes and supporting software system which can help maintainers to understand the system and carried maintenance tasks. Such kind of relation is called traceability links. One approach to discover traceability links is analyzing the similarity of textual content. This paper proposed an approach to discover a traceability links between two software artefacts, which are business processes and software components, using Latent Dirichlet Allocation (LDA). In the proposed method, each label of business process model and software components model are formed into documents. Then, the topic probability distributions are calculated using LDA. The similarities between those two artefacts are calculated using Jensen-Shannon (JS) Divergence The result of LDA is compared to the real software components and business process documents and it shows that LDA and JS Divergence are applicable for discovering traceability links with average Cohens Kappa value of 0.67.