Automatic Uncertainty Detection in Software Architecture Documentation

In this paper, we motivate the usage of natural language processing techniques to detect the uncertainty cues in the software architecture documents. As an initial step of our study, we analyzed three real-world software architecture documents and manually retrieved examples of different types of uncertainties. Based on those examples, we formulated the hypothesis on how the communication of software architecture to its stakeholders could be positively supported by on-time uncertainty detection in software architecture documents. In the end, we outlined our future work and chosen research methodology.