rbofficial Goto Github PK

content_based_recommendation_system

Recommends articles to the user based upon an evaluation of similarity between the content of the article previously consumed by the user and the content of a corpus of 70,000 articles stored in a local NoSQL database. To extract topics from the corpus, three topic modeling techniques- LSA (Latent Semantic Analysis), LDA (Latent Dirichlet Analysis), and NMF (Non-negative Matrix Factorization) are applied and compared. These topics are then used to build the recommendation engine.

handwritten_digit_recognition

personnalise

A Real-Time News Recommendation System that provides a customised news list to each user. Consists of a news pipeline that scrapes news from over ten sources, dedupes and stores them- using MongoDB, Redis and RabbitMQ. Uses Convolutional Neural Network (CNN) to classify the scraped news into 18 predefined categories, and a click log processor to collect user’s click logs for updating users’ preference.

real_time_sentiment_analysis_of_tweets

Identifies the overall sentiment towards a subject of interest by using Tweepy- a Python library to fetch real-time tweets from Twitter, and TextBlob to calculate the sentiment (in terms of polarity values) and subjectivity by parsing each tweet.