This is a replication package for the article Dev2vec: Representing Domain Expertise of Developers in an Embedding Space
The paper aims to encode the expertise of developers, learned from different source of information, into embedding vectors.
These vectors are learned based on doc2vec model that is trained on three different sources of information: repository meta data, issue resolving history and API calls.
We name these models: dev2vec:repos, dev2vec:Issues and dev2vec:APIs
You can download the two models dev2vec:repos and dev2vec:Issues from here
model = Doc2Vec.load("dev2vec_repos")
model = Doc2Vec.load("dev2vec_issues")
The model that is used to generated embedding vectors for dev2vec:APIs is the pretrained model from the article Representation of Developer Expertise in Open Source Software
Dev2vec: Representing Domain Expertise of Developers in an Embedding Space
@article{dakhel2022dev2vec,
title={Dev2vec: Representing Domain Expertise of Developers in an Embedding Space},
author={Dakhel, Arghavan Moradi and Desmarais, Michel C and Khomh, Foutse},
journal={arXiv preprint arXiv:2207.05132},
year={2022}
}