Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SocFedGPT: Federated GPT-based Adaptive Content Filtering System Leveraging User Interactions in Social Networks

Sai Puppala, Ismail Hossain, Md Jahangir Alam, Sajedul Talukder

TL;DR

The paper tackles the challenge of delivering personalized, privacy-preserving content filtering in social networks by integrating context-aware GPT-based models with a federated learning framework. It introduces a User Persona Profiler that combines Engagement, Sentiment, and Readability into per-category scores, and uses these to drive a category-specific content ranking and filtering pipeline. The system employs federated aggregation to train models across distributed clients, a context-based GPT for post categorization, and an adaptive feedback loop to refine recommendations in real time, including readability-based spam filtering. The proposed architecture aims to balance highly personalized content with strong privacy guarantees, enabling scalable, real-time social feed adaptation while safeguarding user data.

Abstract

Our study presents a multifaceted approach to enhancing user interaction and content relevance in social media platforms through a federated learning framework. We introduce personalized GPT and Context-based Social Media LLM models, utilizing federated learning for privacy and security. Four client entities receive a base GPT-2 model and locally collected social media data, with federated aggregation ensuring up-to-date model maintenance. Subsequent modules focus on categorizing user posts, computing user persona scores, and identifying relevant posts from friends' lists. A quantifying social engagement approach, coupled with matrix factorization techniques, facilitates personalized content suggestions in real-time. An adaptive feedback loop and readability score algorithm also enhance the quality and relevance of content presented to users. Our system offers a comprehensive solution to content filtering and recommendation, fostering a tailored and engaging social media experience while safeguarding user privacy.

SocFedGPT: Federated GPT-based Adaptive Content Filtering System Leveraging User Interactions in Social Networks

TL;DR

The paper tackles the challenge of delivering personalized, privacy-preserving content filtering in social networks by integrating context-aware GPT-based models with a federated learning framework. It introduces a User Persona Profiler that combines Engagement, Sentiment, and Readability into per-category scores, and uses these to drive a category-specific content ranking and filtering pipeline. The system employs federated aggregation to train models across distributed clients, a context-based GPT for post categorization, and an adaptive feedback loop to refine recommendations in real time, including readability-based spam filtering. The proposed architecture aims to balance highly personalized content with strong privacy guarantees, enabling scalable, real-time social feed adaptation while safeguarding user data.

Abstract

Our study presents a multifaceted approach to enhancing user interaction and content relevance in social media platforms through a federated learning framework. We introduce personalized GPT and Context-based Social Media LLM models, utilizing federated learning for privacy and security. Four client entities receive a base GPT-2 model and locally collected social media data, with federated aggregation ensuring up-to-date model maintenance. Subsequent modules focus on categorizing user posts, computing user persona scores, and identifying relevant posts from friends' lists. A quantifying social engagement approach, coupled with matrix factorization techniques, facilitates personalized content suggestions in real-time. An adaptive feedback loop and readability score algorithm also enhance the quality and relevance of content presented to users. Our system offers a comprehensive solution to content filtering and recommendation, fostering a tailored and engaging social media experience while safeguarding user privacy.
Paper Structure (18 sections, 8 equations, 3 figures)

This paper contains 18 sections, 8 equations, 3 figures.

Table of Contents

  1. Introduction
  2. Methodologies
  3. Problem statement
  4. Federated Learning
  5. User Persona Profiler
  6. Engagement Score ($E$)
  7. Sentiment Score ($S$)
  8. Category Readability Score ($\rho$)
  9. User Persona Score ($C$)
  10. Content Filtering
  11. Additional Filtering
  12. Readability Score ($R$)
  13. Adaptive Feedback Loop
  14. Experiment
  15. Context GPT
  16. ...and 3 more sections

Figures (3)

  • Figure 1: The system architecture elucidates various components in the Adaptive Content Filtering System approach, with a key focus on GPT, User Persona profile, and Category-based social engagement. Additionally, it provides a concise overview of the distinctive features of the federated learning global server for context-based GPT generation.
  • Figure 2: Continuation of figure 1: The workflow architecture of adaptive content filtering starting from database and all the way to feedback collection from user.
  • Figure 3: User interface for Federated GPT-based Adaptive Content Filtering System Leveraging User Interactions in Social Networks