Document Types
Paper Presentation
School Code
n/a
School Name
De La Salle University Integrated School (Manila)
Abstract/Executive Summary
Social media is a dominant and an ever-expanding platform for social interaction and communication. In its popularity, Social Media Addiction (SMA) emerged as an unintended consequence, affecting a handful of users. Recommender Systems (RS) have been proven to be useful in domains such as health and alcoholism prevention. This study develops a prototype RS against SMA using a user’s personality dimensions and status of SMA. The prototype used a user-based collaborative filtering (CF) approach in recommending items. The prototype will also undergo an evaluation phase but this will not be included in the scope of this paper. The RS prototype accurately predicts an active user’s similarity with his/her neighbors using their scores on TIPI (personality dimensions) and BSMAS (status of SMA) through cosine similarity. Although recommendations show bias towards certain items, they can neither be proven effective nor ineffective without further evaluation such as a User Acceptance Test (UAT). Future prototypes can improve the RS’s information collection phase to reduce inaccuracies.
Keywords
recommender systems; social media addiction; collaborative filtering; personality; cosine similarity
Initial Consent for Publication
yes
A Recommender System against Social Media Addiction among Adolescents
Social media is a dominant and an ever-expanding platform for social interaction and communication. In its popularity, Social Media Addiction (SMA) emerged as an unintended consequence, affecting a handful of users. Recommender Systems (RS) have been proven to be useful in domains such as health and alcoholism prevention. This study develops a prototype RS against SMA using a user’s personality dimensions and status of SMA. The prototype used a user-based collaborative filtering (CF) approach in recommending items. The prototype will also undergo an evaluation phase but this will not be included in the scope of this paper. The RS prototype accurately predicts an active user’s similarity with his/her neighbors using their scores on TIPI (personality dimensions) and BSMAS (status of SMA) through cosine similarity. Although recommendations show bias towards certain items, they can neither be proven effective nor ineffective without further evaluation such as a User Acceptance Test (UAT). Future prototypes can improve the RS’s information collection phase to reduce inaccuracies.
