Detecting rumors in social media using emotion based deep learning approach.

Area of Science:

• Computational Linguistics
• Social Media Analysis
• Artificial Intelligence

Background:

• Social media platforms are primary information dissemination channels.
• Unverified information and rumors on social media can cause significant societal harm.
• Existing rumor detection methods often rely on feature engineering or deep learning without fully exploring emotional and sentimental cues.

Purpose of the Study:

• To develop an advanced deep learning model for accurate rumor detection.
• To investigate the impact of emotional and sentimental analysis on improving rumor detection.
• To propose and validate a novel method that integrates content, emotion, and sentiment for rumor classification.

Main Methods:

• Utilized deep learning approaches to analyze the emotional and sentimental aspects of tweets.
• Developed the Sentiment and EMotion driven TransformEr Classifier (SEMTEC) method.
• SEMTEC extracts emotion and sentiment tags, integrating them with content-based information from the main tweet for semantic analysis.

Main Results:

• Achieved an accuracy rate of 92% for rumor detection on the PHEME dataset.
• Validated the SEMTEC method on a new dataset, Twitter24.
• SEMTEC demonstrated a 2% higher accuracy than standard methods on the Twitter24 dataset.

Conclusions:

• Analyzing user emotional states and sentiment significantly enhances rumor detection capabilities.
• SEMTEC offers a robust and accurate approach to identifying rumors on social media by incorporating multimodal information.
• The proposed method shows promise for real-world applications in combating misinformation.