#70 · Primary category: NLP Tools & Text Processing

neuralcoref

coreference coreference-resolution machine-learning neural-networks nlp python pytorch spacy spacy-extension spacy-pipeline

✨Fast Coreference Resolution in spaCy with Neural Networks

Project last updated:04/13/23

GitHub Stars

2.9K

Forks

469

Contributors

21

License

MIT

Why we included this project

Coreference resolution is the step where a parser figures out that "she" in one sentence points back to a name mentioned earlier. NeuralCoref handles that inside spaCy, as a pipeline extension, so you add it once and then every parsed document carries the resolved coreference clusters. That fits naturally into chatbots, question-answering, and other text-processing systems already built on spaCy, because the annotations stay on the same Document objects you're already working with. It uses a feed-forward neural network to score candidate mention pairs, building on spaCy's existing annotations for mention detection. The main limitation is that the shipped model is pre-trained for English only; other languages need retraining.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category