Affective topic model for social emotion detection.

Area of Science:

• Computational Social Science
• Natural Language Processing
• Affective Computing

Background:

• Social media facilitates widespread emotion communication.
• Traditional sentiment analysis focuses on writer's perspective, which is less relevant for reader emotion detection in social media.
• Bridging the gap between social media content and reader emotions is crucial for understanding online interactions.

Purpose of the Study:

• To propose an affective topic model for detecting emotions evoked in a reader by social media content.
• To enable classification of social emotions in unlabeled documents.
• To generate a social emotion lexicon for enhanced analysis.

Main Methods:

• Development of a novel affective topic model incorporating an intermediate layer.
• Application of the model to classify social emotions in unlabeled social media documents.
• Utilizing real-world social media data for model validation.

Main Results:

• The proposed affective topic model effectively bridges social media content and reader emotions.
• The model demonstrates high accuracy in classifying social emotions of unlabeled documents.
• A social emotion lexicon was successfully generated, proving the model's utility.

Conclusions:

• The affective topic model offers a meaningful approach to reader-centric emotion detection in social media.
• The model is effective for both social emotion classification and lexicon generation.
• This research advances the understanding of reader emotions in the context of social media communication.