Hybrid Recommender System for Mental Illness Detection in Social Media Using Deep Learning Techniques

Area of Science:

• Computer Science
• Artificial Intelligence
• Data Mining

Background:

• Recommender systems are widely used in e-commerce and social media.
• Analyzing user behavior patterns can aid in mental stability assessment.
• Traditional sequential pattern mining algorithms face scalability challenges with large datasets.

Purpose of the Study:

• To develop an optimized recommender system for mental illness detection.
• To improve the efficiency and accuracy of mental health analysis through behavior pattern mining.
• To address the limitations of conventional methods in handling large-scale social media data.

Main Methods:

• Utilized sequential pattern mining for efficient pattern extraction.
• Employed feature selection to remove irrelevant data features.
• Implemented Frequent Pattern (FP) mining with a Systolic tree architecture.
• Developed a hybrid RFD-PSO algorithm for feature selection.

Main Results:

• The Systolic tree architecture offers high throughput and cost-effective performance for FP mining.
• Feature selection enhances classifier speed, accuracy, and cost-effectiveness.
• The proposed recommender system, using hybrid RFD-PSO for feature selection, showed improved performance on depressive patient datasets.

Conclusions:

• Behavior pattern mining combined with advanced feature selection offers a promising approach for mental illness detection.
• The hybrid RFD-PSO algorithm is effective in selecting relevant features for mental health analysis.
• The developed recommender system shows potential for real-world application in mental healthcare.