NLP-Almost All NLP Methods

Sarang Mete
2 min readNov 16, 2022

--

Photo by Hunter Harritt on Unsplash

Assembled almost all NLP concepts in a single use case.

I’ve implemented following NLP techniques individually. You can check their corresponding articles and repo.

  1. Data collection, Data cleaning, Data annotation : collected news data,created nlp text cleaning library and published on Pypi, annotated using doccano
  2. text classification : Classify text into POSITIVE,NEGATIVE,NEUTRAL
  3. Entity coreference resolution
  4. Entity linking,Information extraction,Knowledge graph
  5. Custom NER : CUSTOM_ORG,CUSTOM_PERSON,CUSTOM_PLACE,CUSTOM_ROLE
  6. Searching : Elasticsearch
  7. Code organization, CI/CD

I explored multiple techniques, libraries in each of these projects.

Then, I selected a single best method/model from each of these projects and created a use case where all of them are assembled.

Use Case:

  1. Create an application where user can search for entity(user query) in news articles(knowledge base)
  2. Relevant articles fetched should show sentiment, topics, custom NER entities, knowledge graph
  3. Use best practices in MLOps to develop project. A project quality should be production ready.
Image by Author
Image by Author

I’ve used news articles as knowledge base to search for. You can use any text data(project documents, enterprise documents, web data etc.)

You can refer the github here.

I’ve tried to cover MLOps concepts in each of these projects. Because code quality, CI/CD, testing is very important.

If you liked the article or have any suggestions/comments, please share them below!

Let’s connect and discuss on LinkedIn

--

--