Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SocialRec: User Activity Based Post Weighted Dynamic Personalized Post Recommendation System in Social Media

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

TL;DR

The paper tackles personalized social-media post recommendations by integrating dynamic demographic weights, user post history, and engagement signals to improve relevance and address cold-start and sparsity. It introduces a hybrid framework with weight calculation (demographic-category based weights), traditional matrix factorization, and neural matrix factorization (NeuMF) to model user-item interactions, including a mechanism to share embeddings or concatenate GMF and MLP pathways. Experiments on a synthetic dataset show NeuMF outperforming baselines in Loss, Hit Rate, and NDCG, and demonstrate improved cold-start handling through demographic similarity and cross-user information. The work highlights practical relevance for realistic social feeds and discusses UI implications, privacy concerns, and limitations due to synthetic data, while outlining future work on incorporating temporal dynamics. Overall, the approach advances personalized recommendations by fusing demographic context, historical behavior, and engagement signals into both traditional and neural MF models, with potential impact on cold-start mitigation and feed diversity.

Abstract

User activities can influence their subsequent interactions with a post, generating interest in the user. Typically, users interact with posts from friends by commenting and using reaction emojis, reflecting their level of interest on social media such as Facebook, Twitter, and Reddit. Our objective is to analyze user history over time, including their posts and engagement on various topics. Additionally, we take into account the user's profile, seeking connections between their activities and social media platforms. By integrating user history, engagement, and persona, we aim to assess recommendation scores based on relevant item sharing by Hit Rate (HR) and the quality of the ranking system by Normalized Discounted Cumulative Gain (NDCG), where we achieve the highest for NeuMF 0.80 and 0.6 respectively. Our hybrid approach solves the cold-start problem when there is a new user, for new items cold-start problem will never occur, as we consider the post category values. To improve the performance of the model during cold-start we introduce collaborative filtering by looking for similar users and ranking the users based on the highest similarity scores.

SocialRec: User Activity Based Post Weighted Dynamic Personalized Post Recommendation System in Social Media

TL;DR

The paper tackles personalized social-media post recommendations by integrating dynamic demographic weights, user post history, and engagement signals to improve relevance and address cold-start and sparsity. It introduces a hybrid framework with weight calculation (demographic-category based weights), traditional matrix factorization, and neural matrix factorization (NeuMF) to model user-item interactions, including a mechanism to share embeddings or concatenate GMF and MLP pathways. Experiments on a synthetic dataset show NeuMF outperforming baselines in Loss, Hit Rate, and NDCG, and demonstrate improved cold-start handling through demographic similarity and cross-user information. The work highlights practical relevance for realistic social feeds and discusses UI implications, privacy concerns, and limitations due to synthetic data, while outlining future work on incorporating temporal dynamics. Overall, the approach advances personalized recommendations by fusing demographic context, historical behavior, and engagement signals into both traditional and neural MF models, with potential impact on cold-start mitigation and feed diversity.

Abstract

User activities can influence their subsequent interactions with a post, generating interest in the user. Typically, users interact with posts from friends by commenting and using reaction emojis, reflecting their level of interest on social media such as Facebook, Twitter, and Reddit. Our objective is to analyze user history over time, including their posts and engagement on various topics. Additionally, we take into account the user's profile, seeking connections between their activities and social media platforms. By integrating user history, engagement, and persona, we aim to assess recommendation scores based on relevant item sharing by Hit Rate (HR) and the quality of the ranking system by Normalized Discounted Cumulative Gain (NDCG), where we achieve the highest for NeuMF 0.80 and 0.6 respectively. Our hybrid approach solves the cold-start problem when there is a new user, for new items cold-start problem will never occur, as we consider the post category values. To improve the performance of the model during cold-start we introduce collaborative filtering by looking for similar users and ranking the users based on the highest similarity scores.
Paper Structure (13 sections, 24 equations, 8 figures)

This paper contains 13 sections, 24 equations, 8 figures.

Table of Contents

  1. Introduction
  2. Related Works
  3. Problem Statement
  4. Methodology
  5. Weight Calculation
  6. Matrix Factorization
  7. Neural Matrix Factorization
  8. Experiment
  9. Dataset
  10. Result
  11. Discussion
  12. Application
  13. Conclusion

Figures (8)

  • Figure 1: System Architecture for the Recommendation System with three different approaches
  • Figure 2: How the weight of each demographic attribute is being calculated under the post category.
  • Figure 3: Architecture of Neural Matrix Factorization
  • Figure 4: A demonstration of a recommendation system for three different scenarios by considering (a) users' demographic data and History (b) user engagements or interactions, and (c) result of Hybrid approach
  • Figure 5: Cold start scenario is shown at (a) User feature matrix, (b) Factorization result of User and Post when History and Demographic data are considered as features set, (c) User feature matrix based on engagement, (d) Factorization result of User and Post based on engagement.
  • ...and 3 more figures