Enhancing Twitter Engagement: Content-Based Recommendation Engine Overview
Empower your Twitter experience with a content-based recommendation engine! Discover how the engine analyzes user preferences to suggest relevant profiles to follow, all while increasing engagement and reducing churn. Explore the motivation, solution, similarity algorithms, tech stack, scalability, and future work of this innovative system.
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.
E N D
Presentation Transcript
ParaTweet: A Twitter Content Based Recommendation Engine Sanketh Katta Anthony Salgado Rohit Turumella Jamie Turley Mentor : Shai Haim
Motivation Lots of users on Twitter but it s hard to find new users that are similar to people I already follow Similar to You Feature Current approaches are centered around the concept of triadic closure Follow and List Recommendations help generate growth, increase user engagement, and reduces churn
Our Solution Generate recommendations based on the content that a given user consumes This allows us to do textual analysis to get a better idea of the type of content that the user prefers Web Application that allows users to enter a Twitter Handle and get a list of personalized recommendations for users that they should follow
Similarity Algorithm Other approaches include TF-IDF, Jaccard Coefficient Noticed better performance and recommendations with Cosine Similarity Easy to implement hard to compute with given vector size Recommendation Engine is meant not to run in real time (background process)
The Stack Frontend: Twitter Bootstrap, JQuery Backend: Python-Flask, MongoDB, Scaling: Green Unicorn, Nginx, Supervisor Scraping: Python-Twitter, Stemming (Porter Stemming)
How would this work at Twitter scale? Every user would have pre-generated recommendations which are periodically refreshed The algorithm would be run as a Cron Job Ability to Parallelize Algorithm with access to a cluster Currently running on one machine Not limited by the REST API Getting the data through the REST API was a bottleneck
Future Work Parallelize and Optimize Scraping and Recommendation Algorithms Breadth-First Traversal of all Twitter users to generate reports on the fly Introduce support for foreign language users Enhance corpus of data (ground truth users) with notable individuals in different countries Add a signal to facilitate location-based recommendations
