Overview

This repo explores rapid-prototyping extractive/abstractive text summarization methods. Primary method used for extractive summarization is cosine similarity and ranking. Currently the application is wrapped in a REST API, plans for next features includes wrapping in protobuf/grpc system for much faster api calls, using pre-trained word2vec systems, and caching.

Demo:

curl -d '{"url":"https://www.theatlantic.com/health/archive/2020/03/how-will-coronavirus-end/608719/"}' -H "Content-Type: application/json" -X POST https://nlprnd.com/api/v1/getsummary

Dependencies:

• Docker
• Docker compose
• python 3

Intructions to run:

1. docker-compose build
2. docker-compose up
3. App running on port http://0.0.0.0:80/
4. POST to http://0.0.0.0:80/api/v1/getsummary using: { "text": "Text you wish to summarize" } or { "url": "Url of any article" }

Changenotes

• updated reqs