ContextsFrench

ContextsFrench is an open‑source software library designed to provide context‑aware natural language processing tools specifically tailored for the French language. Developed in 2018 by a consortium of researchers from several European universities, the library builds upon the popular spaCy framework, extending its core capabilities with modules for morphological disambiguation, syntactic parsing, and semantic role labeling that account for the unique features of French grammar. The project’s main goal is to improve the accuracy of downstream applications such as question answering, machine translation and sentiment analysis when dealing with complex French text, including literary prose, legal documents and conversational dialogue.

The library offers a set of pre‑trained contextual word embeddings based on BERT and XLM‑Roberta, fine‑tuned on large French corpora such as the French Wikipedia, the OpenSubtitles dataset, and the Corpus de Référence. In addition to embeddings, ContextsFrench supplies rule‑based post‑processing routines that handle French clitic pronouns, elision and gender agreement. The API is intentionally lightweight, exposing a single `ContextProcessor` class that accepts raw text, returns parsed tokens and an optional semantic graph. All models and training scripts are released under an MIT license, enabling developers to incorporate the library into proprietary software.

ContextsFrench has been cited in several peer‑reviewed publications on multilingual natural language understanding, and it has been integrated into leading data‑science platforms to assist enterprises dealing with French‑speaking market data. The project maintains an active GitHub repository where contributors can report issues, submit pull requests, and discuss improvements with the community.