Design considerations for supporting social interaction in algorithmic social media feed

Abstract

Nowadays, social media platforms employ recommender systems to proactively suggest the posts that a user is likely to prefer, even if it comes from strangers (e.g., the “Suggested Posts” feature on Instagram). Previous studies revealed that these algorithmic recommendations influence users’ social interaction practices in social media. However, there is little attempt to investigate how social media feeds should be designed to ensure the social interactions users truly want. Given this background, we conducted four design workshops aiming to explore design considerations for social media feeds that can support users’ social interactions under the impact of algorithmic recommendations. Based on the results of workshops, we present five design considerations with design proposals: respecting the following posts, preventing undesired disclosure, sophisticated management for recommendation algorithms, recommendations for the group of users, and reinforcing linkage to the community. In conclusion, we discuss design implications for recommender systems considering the intrinsic nature of social media.