Modeling of Recommendation System Based on Emotional Information and Collaborative Filtering

Area of Science:

• Speech emotion recognition
• Affective computing
• Recommender systems

Background:

• Current recommendation systems lack personalization, failing to account for user emotions and preferences.
• User satisfaction is limited by the inability of existing methods to accurately reflect individual emotional states.
• Emotion information is crucial for tailoring services, such as music recommendations and user monitoring.

Purpose of the Study:

• To develop a system that accurately recognizes user emotions from speech.
• To classify content based on emotional attributes.
• To enhance content recommendation by matching user emotions with suitable content.

Main Methods:

• Utilized Genetic Algorithms as a Feature Selection (GAFS) method for speech normalization and classification.
• Employed a Support Vector Machine (SVM) algorithm with kernel function optimization for emotion recognition.
• Applied factor analysis, correspondence analysis, and Euclidean distance for content classification based on emotion.
• Integrated collaborative filtering to predict user emotional preferences.

Main Results:

• Achieved a high emotion recognition accuracy of 86.98% using the Radial Basis Function (RBF) kernel with SVM.
• Successfully classified content (images, music) according to recognized emotional information.
• Developed a mobile application capable of recommending content aligned with user emotions.

Conclusions:

• Speech emotion recognition is a viable method for enhancing personalized content recommendations.
• The proposed system effectively bridges the gap between user emotional states and content suitability.
• This approach has the potential to significantly increase user satisfaction in various applications.