GROUP MEMBERS:
- Akul V Jha - 21MA10008
- Saksham Gupta - 21MA10051
- Yash Garg - 21MA10066
- Dheeraj Chandak - 21MA10021
- Anshul Agarwal - 21MA10010
- Kapil Kumar - 21MA10026
PROBLEM STATEMENT: In the era of social media dominance, the significance of gauging the popularity of content, particularly on platforms like Twitter, has become increasingly vital. As users generate an incessant stream of tweets, discerning which will gain popularity poses a challenge. This project aims to develop a predictive model that can analyze Twitter tweets and forecast their potential popularity, providing valuable insights into the factors contributing to the virality of content on the platform.
The challenges encompassed in this endeavour include the dynamic nature of online conversations, the diverse content types present in tweets, and the ever-evolving preferences of the user base. Understanding the dynamics of popularity in this context involves exploring features such as engagement metrics (likes, retweets, replies), textual content analysis, temporal patterns, and user interactions. By delving into these facets, we seek to create a robust model that distinguishes between tweets that will gain traction and those that may not.
The project's success will not only contribute to a deeper understanding of social media dynamics but will also have practical implications for individuals and organizations seeking to enhance their online presence and impact. Consequently, this research aims to bridge the gap between the plethora of user-generated content and the ability to identify and harness the potential for virality in the context of Twitter.
CONCLUSION: Our project revolves around predicting content popularity on Twitter and YouTube using the Hawkes Intensity Processes (HIP) model applied to the rich ACTIVE dataset. HIP's effectiveness is demonstrated through parameter fitting and model evaluation, showcasing its ability to capture the interplay between inherent video appeal and external promotions.
Extending predictions from day 91 to day 120, our forecasting phase equips content creators with a valuable tool for optimizing strategies and anticipating engagement fluctuations.
As we conclude, future avenues include real-time implementation, dynamic feature incorporation, multimodal data integration, and ethical considerations, ensuring our project remains at the forefront of social media analytics. This project lays the groundwork for ongoing inquiry into the evolving landscape of online content popularity, leveraging sophisticated mathematical modeling, data exploration, and predictive analytics for insightful analyses in the realm of social media dynamics.