Title:	Analysis of factors affecting user engagement patterns with different classes of news on social media
DNr:	Berzelius-2023-367
Project Type:	LiU Berzelius
Principal Investigator:	Alireza Mohammadinodooshan <alireza.mohammadinodooshan@liu.se>
Affiliation:	Linköpings universitet
Duration:	2023-12-19 – 2024-03-01
Classification:	10204
Keywords:

Abstract

Objective: With this analysis, we aim to study the factors influencing user engagement with news content on Social media. We have compiled an expansive 152 million tweets and Facebook posts dataset spanning over 5 years from more than 2,280 news publishers. All the posts are augmented with the engagement data. Employing state-of-the-art natural language and image processing models, we will quantify how attributes like tweet sentiment, emotions, topics, image features, publisher ideology, and content reliability correlate with engagement rates. Significance: This research will advance academic comprehension of how various post properties interact with publisher attributes to shape engagement patterns. Findings can enlighten content strategies for publishers and platforms regarding fostering healthier discourse. Moreover, our large-scale tweet annotation methodology demonstrates the value of leveraging impressions data and deep tweet embeddings to clarify social media dynamics.