The reason for that request was just that I use a BibTex file where numbers are needed for other applications. And I wanted to avoid additionnal file preprocessing for bibbase. But I understand your point. I'll do my own changes.

I see. However, isn't it possible for BibBase to manage with the "source_type" or the "user_context" field ? As you can see, it gives : 

 source_type = {mastersthesis},
 user_context = {Master's thesis},

on (previously) https://bibbase.org/service/mendeley/fdd686a8-32de-38c0-97f8-853ef3fffb3e:

@phdthesis{
 title = {Robustness for GOssip Private Averaging},
 type = {phdthesis},
 year = {2018},
 keywords = {Byzantine attack,Paillier homomorphic encryption,churn tolerance,decentralized protocols,gossip averaging,privacy,zero-knowledge proofs},
 pages = {1-72},
 websites = {https://alexandrehuat.github.io/Huat2018-MastersThesis-RobustGOPA.pdf},
 month = {11},
 city = {685 Avenue de l'Université, 76800 Saint-Étienne-du-Rouvray, France},
 institution = {INSA Rouen Normandie and University of Rouen Normandie},
 department = {Information Systems Architectures, Data Science},
 id = {d1bb7354-0962-3625-bb39-03031377a86f},
 created = {2019-03-14T08:19:21.373Z},
 file_attached = {false},
 profile_id = {fdd686a8-32de-38c0-97f8-853ef3fffb3e},
 last_modified = {2019-03-14T12:53:00.915Z},
 read = {false},
 starred = {false},
 authored = {true},
 confirmed = {true},
 hidden = {false},
 citation_key = {Huat2018},
 source_type = {mastersthesis},
 user_context = {Master's thesis},
 private_publication = {false},
 abstract = {GOPA is a privacy-preserving gossip algorithm for averaging that has applications in decentralized machine learning over P2P networks. This Master's thesis aims to improve the robustness of the algorithm to churns and malicious adversaries. Thanks to the Paillier cryptosystem, commitment schemes and mathematical analysis, we have convincing theoretical results regarding both privacy aspects and correctness of averaging. Notable contributions are a framework for churn-tolerance and a privacypreserving algorithm for detecting data injections in gossip averaging. In addition, we unify all our improvements around a robust user-centric version of GOPA. All our contributions have proved to be quite novel in comparison with the literature. We also have initiated the development of an Android prototype of GOPA so that our solutions can be experimented in the future. Finally, new theoretical and technical issues have been identified and may be treated as part of a next internship or PhD studies.},
 bibtype = {phdthesis},
 author = {Huat, Alexandre}
}

Actually, I've already had problems with my self-hosted BibTeX file. That's why I tried with Mendeley. Anyway, I'll see what I can do and will open a new issue if my self-hosted file still bugs.

Thank you.