Nepal, Surya, et al. "Interaction-based recommendations for online communities." ACM Transactions on Internet Technology (TOIT) 15.2 (2015): 1-21.

This article explores the challenge of keeping online communities aimed at supporting behavioural change (such as those focused on diet, lifestyle, or the environment) active and engaged. Unlike traditional social networks that rely on existing friendships for engagement, these communities lack dense friendship networks. To address this, the article introduces a social behaviour model applied to both people and content recommendation systems, designed to foster connections based on member interactions and shared interests rather than pre-existing friendships. An evaluation conducted within an online diet and lifestyle community of 5,000 Australians participating in a 12-week program shows that this approach significantly outperforms traditional recommendation algorithms based on friendships or link predictions.