Query Expansion Based On User Requirements Clustering for Finding Feature Location

Feature Location is an approach of how to determine the code area based on high level software artifacts. Use Case Scenarios (UCS) are requirements documents in software artifacts that contain many words. A sentence in a UCS is sometimes described by a sentence in other UCS, so capture the relationship among UCS may be advantageous in finding feature locations. The research contribution is how to find feature locations better by making an expansion to query based on UCS clustering. The method was clustering the UCS using k-medoids clustering and index the source code using latent dirichlet allocation. The results was 56.8% for the best recall rate.