Predicting Viral Memes & Internet Trends
Michael Casarona
Co-Presenters: Lark Bancairen
College: The Dorothy and George Hennings College of Science, Mathematics and Technology
Major: Computational Science & Engineering - STEM 5 Year B.S./M.S.
Faculty Research Mentor: Ching-yu Huang
Abstract:
The goal of this research is to develop a predictive model that identifies viral memes and internet trends before they peak in popularity. With the rapid evolution of digital culture, viral content plays a crucial role in shaping online interactions, marketing strategies, digital investments, and social engagement. By analyzing key factors influencing meme virality, this project aims to provide valuable insights for businesses, content creators, and analysts to capitalize on emerging trends.For this study, large-scale datasets were collected and analyzed from various platforms (e.g. Twitter, Reddit, Google Trends, etc.). The datasets contain a diverse range of attributes, including textual data (hashtags, captions, engagement metrics), media features (image/video metadata), and sentiment scores. The data underwent preprocessing through an ETL pipeline, involving multiple steps to clean and refine the data for analysis. Initially, raw data was extracted from structured and unstructured sources using APIs and dataset repositories. During transformation, the data was cleaned and natural language processing (NLP) was applied for sentiment analysis. Additionally, new features were engineered, such as virality scores (based on engagement metrics) and meme category classification using deep learning. Finally, the cleaned and structured data was loaded into a relational database for efficient querying and model training.Preliminary findings suggest that meme virality is influenced by engagement levels, content relatability, and platform-specific dynamics. While further model optimization is ongoing, early results indicate that predictive modeling can effectively forecast internet trends. This research contributes to a deeper understanding of digital virality and its broader implications for content strategy and trend analysis.