No abstract

Abstract-We posit that machine learning can be applied to effectively address requirements engineering problems. S pecifically, we present a requirements traceability method based on the machine learning technique Reinforcement Learning (RL). The RL method demonstrates a rather targeted generation of candidate links between textual requirements artifacts (high level requirements traced to low level requirements, for example). The technique has been validated using two real -world datasets from two problem domains.Our technique demonstrated statistically significant better results than the Information Retrieval technique.

Abstract-We posit that swarm intelligence can be applied to effectively address requirements engineering problems. Specifically, this paper demonstrates the applicability of swarm intelligence to the requirements tracing problem using a simple ant colony algorithm. The technique has been validated using two real-world datasets from two problem domains. The technique can generate requirements traceability matrices (RTMs) between textual requirements artifacts (high level requirements traced to low level requirements, for example) with equivalent or better accuracy than traditional information retrieval techniques.

We present the process and methods applied in undertaking the Traceability Challenge in addressing Grand Challenge C-GC1 -Trace recovery. The Information Retrieval methods implemented in REquirementsTRacing On target .NET (RETRO.NET) were applied to the tracing of the eTour and EasyClinic datasets. Our work focused on the nuances of native language (Italian, English). Datasets were augmented with additional terms derived from splitting function and variable names with Camel-Back notation and using the Google Translate API to translate Italian terms into English. Results based on the provided answer set show that the augmented datasets significantly improved recall and precision for one of the datasets.

No abstract