Human Micro-Expressions in Multimodal Social Behavioral Biometrics

Area of Science:

• Computer Science
• Cybersecurity
• Human-Computer Interaction

Background:

• Social Behavioral Biometrics (SBB) are crucial for online person identification.
• Understanding unique social interaction and communication patterns is key.
• Existing SBB systems can be enhanced with novel biometric traits.

Purpose of the Study:

• To introduce a novel multimodal SBB system.
• To integrate human micro-expressions from text as a new biometric trait.
• To enhance online user identification performance.

Main Methods:

• Developed a multimodal SBB system incorporating human micro-expressions.
• Extracted six SBB traits, including micro-expressions, for comprehensive user representation.
• Employed rank-level fusion using weighted Borda count to combine trait scores.

Main Results:

• The proposed method achieved 73.87% accuracy and 74% recall on a Twitter user dataset.
• Incorporating human micro-expressions substantially boosted identification performance.
• Outperformed existing state-of-the-art SBB systems.

Conclusions:

• Human micro-expressions are a valuable biometric trait for SBB.
• Multimodal SBB systems integrating diverse traits offer superior performance.
• The proposed fusion method effectively enhances online user identification.