Tap-news
A Real Time News Scraping And Recommendation System
Run
cd into client npm install npm run build
cd into server npm start
./news_pipeline_launcher.sh
Used
Front-end: React
Back-end: python RPCServer, redis, express, mongodb, RabbitMQ
use redis to check if it's already scrapped
use watchdog to observe the update in the model
use RabbitMQ to fetch and send news message
Model: NLP, DNN, Tensorflow
use to predict the topic
update user's preference from their click events by using Time decay model p = (1-α)p + α & p = (1-α)p
use pairwise_sim = tfidf * tfidf.T to de-duplicate similar news