This version of trace2vary have achieved better results using Stochastic Gradient Descent and AutoML machine learning methods, if compared to the traditional Information Retrieval (IR) models: classic vector, extended boolean, BM25 and neural networks.
Descriptive results are available in the 'trace_recovery/results' folder.
In addition the dataset used to validate the whole pipeline was the set of four BSD operating system products:
projects/dragonflybsd
projects/freebsd
projects/netbsd
projects/openbsd
The results for such projects are available in the 'output/trace2vary_result.html' file.