TheConversationSentimentAnalysis

This is just a sample project built around Quart to do summarization and sentiment analysis of news articles from The Conversation. The Conversation uses a Creative Commons NoDerivatives license. Please note, that this site will not be officially hosted as summarization may fall under a 'remix' or 'transform' of the original material.

I wanted to learn several new things. Here's what I picked.

I chose to use Quart for this. Quart is an async reimplementation of Flask. This is useful, especially for things like inference which may take some time.

I also utilized websockets to handle sending of information.

I picked tox to help automate with formatting and test checks.

I used huggingface to pick a llm model to help do summarization.

This is not a showcase of styling, or ultra performant web front-end.

To run locally:

cd app hypercorn --reload --bind 0.0.0.0:5000 --workers 1 main:app

I utilize tox for formatting and cleanliness. To run, go to the root of the repo and run 'tox' to run the tests/formatting.