Trace Link Recovery Using Semantic Relation Graphs and Spreading Activation

The paper was first published at the 28th IEEE International Requirements Engineering Conference in 2020. Trace Link Recovery tries to identify and link related existing requirements with each other to support further engineering tasks. Existing approaches are mainly based on algebraic Information Retrieval or machine-learning. Machine-learning approaches usually demand reasonably large and labeled datasets to train. Algebraic Information Retrieval approaches like distance between tf-idf scores also work on smaller datasets without training but are limited in providing explanations for trace links. In this work, we present a Trace Link Recovery approach that is based on an explicit representation of the content of requirements as a semantic relation graph and uses Spreading Activation to answer trace queries over this graph. Our approach is fully automated including an NLP pipeline to transform unrestricted natural language requirements into a graph. We evaluate our approach on five common datasets. Depending on the selected configuration, the predictive power strongly varies. With the best tested configuration, the approach achieves a mean average precision of 40% and a Lag of 50%. Even though the predictive power of our approach does not outperform state-of-the-art approaches, we think that an explicit knowledge representation is an interesting artifact to explore in Trace Link Recovery approaches to generate explanations and refine results