NLP-guidance

Natural Language Processing guidance

This repo contains thoughts and guidance about the use of Natural Language Processing, based on the experience of using these techniques in a few projects here at the Data Science Hub in the Ministry of Justice.

If you have any thoughts or feedback on any of this, especially if you disagree, please email me.

Contents

Introduction

Things we have done with NLP:

• Search through a set of documents
• Find topics in a set of documents

Techniques we have used

• Feature Selection
• Latent Semantic Analysis (LSA)
• Latent Dirichlet Allocation (LDA)
• Word2Vec, Doc2Vec, fastText (Neural Network models)

Glossary

Code

• There is code so you can try out these techniques in practice, located in the code folder in the repo: https://github.com/moj-analytical-services/NLP-guidance
• More details about it here