Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Amplifying Your Social Media Presence: Personalized Influential Content Generation with LLMs

Yuying Zhao, Yu Wang, Xueqi Cheng, Anne Marie Tumlin, Yunchao Liu, Damin Xia, Meng Jiang, Tyler Derr

TL;DR

This work investigates whether LLMs can generate personalized content that meaningfully amplifies a user’s social-media presence. It introduces a content-aware influence estimator and develops content-centric and structure-aware prompting strategies to maximize diffusion in a social network. Across Weibo experiments with Phi-2, Phi-3, and Mistral, the authors demonstrate consistent improvements in both local repost probabilities and global influence spread, with structure-aware prompts providing notable benefits. The study highlights the importance of network neighborhood information, estimator accuracy, and user-group dynamics, while acknowledging ethical considerations and the need for safeguards against misinformation.

Abstract

The remarkable advancements in Large Language Models (LLMs) have revolutionized the content generation process in social media, offering significant convenience in writing tasks. However, existing applications, such as sentence completion and fluency enhancement, do not fully address the complex challenges in real-world social media contexts. A prevalent goal among social media users is to increase the visibility and influence of their posts. This paper, therefore, delves into the compelling question: Can LLMs generate personalized influential content to amplify a user's presence on social media? We begin by examining prevalent techniques in content generation to assess their impact on post influence. Acknowledging the critical impact of underlying network structures in social media, which are instrumental in initiating content cascades and highly related to the influence/popularity of a post, we then inject network information into prompt for content generation to boost the post's influence. We design multiple content-centric and structure-aware prompts. The empirical experiments across LLMs validate their ability in improving the influence and draw insights on which strategies are more effective. Our code is available at https://github.com/YuyingZhao/LLM-influence-amplifier.

Amplifying Your Social Media Presence: Personalized Influential Content Generation with LLMs

TL;DR

This work investigates whether LLMs can generate personalized content that meaningfully amplifies a user’s social-media presence. It introduces a content-aware influence estimator and develops content-centric and structure-aware prompting strategies to maximize diffusion in a social network. Across Weibo experiments with Phi-2, Phi-3, and Mistral, the authors demonstrate consistent improvements in both local repost probabilities and global influence spread, with structure-aware prompts providing notable benefits. The study highlights the importance of network neighborhood information, estimator accuracy, and user-group dynamics, while acknowledging ethical considerations and the need for safeguards against misinformation.

Abstract

The remarkable advancements in Large Language Models (LLMs) have revolutionized the content generation process in social media, offering significant convenience in writing tasks. However, existing applications, such as sentence completion and fluency enhancement, do not fully address the complex challenges in real-world social media contexts. A prevalent goal among social media users is to increase the visibility and influence of their posts. This paper, therefore, delves into the compelling question: Can LLMs generate personalized influential content to amplify a user's presence on social media? We begin by examining prevalent techniques in content generation to assess their impact on post influence. Acknowledging the critical impact of underlying network structures in social media, which are instrumental in initiating content cascades and highly related to the influence/popularity of a post, we then inject network information into prompt for content generation to boost the post's influence. We design multiple content-centric and structure-aware prompts. The empirical experiments across LLMs validate their ability in improving the influence and draw insights on which strategies are more effective. Our code is available at https://github.com/YuyingZhao/LLM-influence-amplifier.
Paper Structure (29 sections, 5 equations, 9 figures, 2 tables)

This paper contains 29 sections, 5 equations, 9 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Influence Estimation
  4. LLMs in Social Media
  5. Content-Aware Influence Estimator
  6. Pairwise Influence Estimation
  7. Content Influence Estimation
  8. Problem Statement
  9. Content-Centric Strategies
  10. Structure-Aware Strategies
  11. Representative Posts
  12. Summarized Interests
  13. Experiments
  14. Settings
  15. RQ1: Effectiveness
  16. ...and 14 more sections

Figures (9)

  • Figure 1: Many social media users strive to maximize their influence, seeking more reposts and likes. In reality this is quite challenging and they may only receive engagement from a small portion of their audience. How can a user create a post that gains larger influence?
  • Figure 2: Online community structure matters: the same post related to LLM will spread differently in a tech-friendly versus tech-skeptical local neighborhood.
  • Figure 3: The overall framework consists of the following steps: (1) input data are collected based on prompt requirement; (2) prompts are generated using content-centric and structure-aware strategies; (3) the prompts are fed into LLMs to generate posts; and (4) the influence of the generated posts is evaluated through spread modeling.
  • Figure 4: Probability comparison between the original post and the revised post.
  • Figure 5: Influence evaluation performance comparison.
  • ...and 4 more figures