Integrating Deep Learning and Expert Systems in Negotiation Support: A Framework for Body Language Recognition to Augment Decision-Making Via Non-Verbal Cues

This study develops an AI-driven expert system to improve negotiation outcomes by analyzing body language, a key but often ignored aspect of human communication in decision-making. The proposed system integrates Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks to serve as a knowledge-driven inference engine. By analyzing skeletal keypoint data, it classifies ten negotiation-specific behavioral cues, including weak stance, strong stance, overconfidence, interest, and openness. This approach enhances traditional rule-based negotiation models by augmenting them with data-driven insights, demonstrating the system's potential to improve decision-making processes. Translating body movements into behavioral insights mimics human expertise to help resolve conflicts or biases during group discussions. A custom dataset, created with input from negotiation experts and using skeletal data, supports the model. Early tests show 99% accuracy, proving its potential to enhance traditional negotiation tools with real-time, adaptive advice. However, challenges like cultural differences in interpreting gestures and data biases need addressing. The framework integrates into Negotiation Support Systems (NSS) as a ``smart advisor'' that learns over time and explains its reasoning, building user trust. Blending technical precision with psychological understanding advances AI systems that better grasp human interaction. Future steps include adding voice or eye-tracking data and testing the system across diverse cultural settings to ensure broader applicability in high-stakes scenarios. This work bridges technology and human behavior, aiming to create AI tools that foster collaboration in complex negotiations.