Skip to content

A web app for a simple named-entity recognition system

Notifications You must be signed in to change notification settings

qdgiang/ner-system

Repository files navigation

NER System

A web app for a simple named-entity recognition system

Model

There are three models in this web app, two trained by me using Conditional Random Field, and one using spacy python package.

I have also trained another model using Distributed Random Forest with 92% accuracy, but the model can only be exported in Java, and sklearn Random Forest, but the model only reached 42% so lower than the CRF which has an accuracy of 85% when we disregard the O tag.

The two models are for Formal and Casual sentence.

  • Formal sentence means sentence with appropiate capitalization and correct English words and grammar.

  • Casual sentence means sentence without correct capitalization and such.

Web deployment

View the web app here: https://ner-simple-system.herokuapp.com/

About

A web app for a simple named-entity recognition system

Resources

Code of conduct

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published