This repo contains example projects for various NLP tasks, implemented with spaCy and other NLP frameworks. The projects include scripts, benchmarks and results, as well as annotated datasets (created with Prodigy).
| Name | Description |
|---|---|
ner-food-ingredients |
Use sense2vec and Prodigy to bootstrap an NER model to detect ingredients in Reddit comments and to calculate how these mentions change over time. Includes an end-to-end video tutorial, raw pre-processed data, 949 annotated examples and pretrained tok2vec weights. |
ner-fashion-brands |
Use sense2vec to bootstrap an NER model to detect fashion brands in Reddit comments. Includes 1735 annotated examples, a data visualizer, training and evaluation scripts for spaCy and pretrained tok2vec weights. |
ner-drugs |
Use word vectors to bootstrap an NER model to detect drug names in Reddit comments. Includes 1977 annotated examples, a data visualizer, training and evaluation scripts for spaCy and pretrained tok2vec weights. |
textcat-docs-issues |
Train a binary text classifier with exclusive classes to predict whether a GitHub issue title is about documentation. Includes 1161 annotated examples, a live demo and downloadable model and training and evaluation scripts for spaCy. |
nel-emerson |
Use spaCy and Prodigy to train an Entity Linking model to disambiguate mentions of "Emerson" to unique WikiData identifiers. |