Abstract
In online discussion forums, users are more motivated to take part in discussions when observing other users’ participation—the effect of social influence among forum users. In this paper, we study how to utilize social influence for increasing the overall forum participation. To this end, we propose a mechanism to maximize user influence and boost participation by displaying forum threads to users. We formally define the participation maximization problem, and show that it is a special instance of the social welfare maximization problem with submodular utility functions and it is NP-hard. However, generic approximation algorithms is impracticable for real-world forums due to time complexity. Thus we design a heuristic algorithm, named Thread Allocation Based on Influence (TABI), to tackle the problem. Through extensive experiments using a dataset from a real-world online forum, we demonstrate that TABI consistently outperforms all other algorithms in maximizing participation. The results of this work demonstrates that current recommender systems can be made more effective by considering future influence propagations. The problem of participation maximization based on influence also opens a new direction in the study of social influence.
Introduction
Online discussion forums thrive on active participation—more posts generate more value for users and platform owners. Traditionally, forum recommendation engines suggest threads based on a user's historical interests. However, this "isolated" approach ignores the social ecosystem of the forum. Users are social creatures: they are more likely to post when they see others participating. If a system recommends a thread to an "influencer" who then posts, that single post might trigger several other users to join the conversation. This paper introduces the participation maximization problem, shifting the focus from simple interest matching to strategically leveraging social influence to maximize total forum activity.
Methodology
User Posting Model
The researchers define a stochastic user posting model based on the Independent Cascade model. In this framework, a user reads a thread and is "triggered" to post with a certain probability based on the influence of previous posters.
The TABI Algorithm
To solve the maximization problem, which is proven to be NP-hard, the authors introduce TABI (Thread Allocation Based on Influence). Unlike standard recommenders, TABI’s scoring function considers two factors:
- Interest/Past Influence: The likelihood of a user being influenced by existing posts.
- Future Influence: The likelihood of this user triggering other inactive users if they post.
TABI uses a "fast neighborhood calculation" rather than expensive simulations, making it practical for real-time systems.
Datasets
The approach was evaluated using real-world data from TripAdvisor’s World travel forum, specifically focusing on threads and users from Orlando, London, and New York City. An implicit social influence network was constructed by analyzing posting sequences across thousands of threads.
Results and Discussion
The authors prove that the participation function is "submodular," meaning it exhibits diminishing returns. This property allows for efficient approximation. In head-to-head simulations, TABI consistently outperformed baseline methods, random allocation, and state-of-the-art personalized recommendation algorithms like TEABIF.
Key findings include:
- Performance Boost: TABI achieved a 5.5% to 6.2% increase in new participants compared to standard personalized recommendations.
- Baselines: Compared to no sidebar, TABI increased participation by 50-60%.
- Efficiency: TABI provides high-quality estimates almost instantly by looking at immediate neighbors in the influence graph, whereas theoretical algorithms like RPA require thousands of simulations.
Conclusion
This work demonstrates that recommendation systems can be significantly improved by accounting for future influence propagations. While the current model simplifies user interactions (e.g., ignoring negative interactions or content quality), it establishes a robust foundation for influence-based participation maximization. Future work will involve live user studies and refining the model to account for deeper influence cascades.
Summary updated