Digital media recommendation system design based on user behavior analysis and emotional feature extraction

Area of Science:

• Computer Science
• Artificial Intelligence
• Data Science

Background:

• Digital platforms generate vast user sentiment data crucial for understanding behavior.
• Current methods for analyzing this data and tailoring content are limited.
• User-centric emotional insights are vital for optimizing content strategies and user satisfaction.

Purpose of the Study:

• To introduce the ATLSTM-PS framework for content dissemination strategies based on user emotions.
• To enhance feature extraction precision in sentiment analysis.
• To improve content recommendation accuracy and user experience on digital platforms.

Main Methods:

• Utilized an enhanced ATT-LSTM method with an attention mechanism for precise emotional content extraction.
• Integrated user behavioral and emotional attributes at the feature layer.
• Developed the ATLSTM-PS framework to synthesize these integrated features for accurate recommendations.

Main Results:

• The ATLSTM-PS framework demonstrated significantly enhanced efficacy in content dissemination.
• The synergy of distinct attention layers in ATLSTM-PS improved recommendation precision.
• Empirical results validated the framework's effectiveness on diverse datasets.

Conclusions:

• ATLSTM-PS offers a novel technical tool for sentiment analysis in digital media.
• The framework provides a potent methodology for multimedia platforms to refine information dissemination.
• This approach leads to augmented user experiences through personalized content recommendations.