Content and Context Based Analysis of University Social Media Engagement

Muhammad Farrel,Ade Romadhony,Hasmawati Hasmawati
DOI: https://doi.org/10.35957/jatisi.v9i4.2696
2022-12-13
JATISI (Jurnal Teknik Informatika dan Sistem Informasi)
Abstract:Social media has become an essential tool for institutions to create interactions with their users. Creating successful and relevant material for social media content is challenging as institutions struggle to comprehend what drives user engagement. Using analytics on user-generated social media material to determine the relation between post features and user engagement becomes a method for overcoming this challenge. In this study, engagement is measured based on a quantitative approach to the number of likes and comments. We applied machine learning algorithm to extract contextual features (time of post) and content features (caption and image classification) from university Instagram posts, and their influence on user engagement is statistically modeled. LASSO regression is applied, which is an effective strategy for automated variable selection. The results indicate that context features around 13:00 on Sunday, during the month of June, and content features containing the words "Gedung" and "Foto," as well as image classifications of "Sight" and "Sadness," will receive more likes and comments.
What problem does this paper attempt to address?