A model for the Twitter sentiment curve.

Area of Science:

• Computational Social Science
• Network Science
• Sentiment Analysis

Background:

• Twitter is a primary platform for political communication due to its message concision and rapid diffusion.
• Understanding tweet sentiment is vital for predicting discussion evolution and narrative trends, especially for emotionally charged topics.

Purpose of the Study:

• To present a novel computational model for reproducing and predicting the dynamics of tweet sentiments on Twitter.
• To analyze the emotional sensitivity of public responses to events based on Twitter data.

Main Methods:

• Utilized a recent variant of the Pólya urn model with local reinforcement and random persistent fluctuation.
• Applied the model to various Twitter datasets to evaluate its performance in sentiment prediction.
• Compared the proposed model against the standard Pólya urn model.

Main Results:

• The proposed Pólya urn model demonstrated superior performance in reproducing and predicting tweet sentiment dynamics compared to the standard model.
• The model successfully captured trend fluctuations in sentiment curves, reflecting real-time user emotional responses.
• Performance variations across datasets indicated differing public emotional sensitivities to specific events.

Conclusions:

• The developed Pólya urn model offers a robust tool for analyzing and forecasting sentiment in online political discourse.
• The model's ability to capture sentiment fluctuations provides insights into the real-time impact of public events on social media narratives.
• This approach has broad applicability beyond political science, enabling sentiment analysis in diverse online communication contexts.