Reinforcement Learning Agents for Automated Contract Negotiation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing negotiation automation systems are limited in training agents with diverse behavioral patterns for contract negotiation, reducing efficiency and scalability, and require extensive training data from multiple domains to imitate human behavior effectively.
Innovation Solution
A system and method using reinforcement learning agents with multiple behavioral models, including Selfish-Selfish, Selfish-Prosocial, Prosocial-Selfish, and Prosocial-Prosocial models, to negotiate and select optimal contract proposals through a negotiation training procedure, where a selector agent chooses the best proposal based on a reward function, enabling efficient and scalable negotiation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional face-to-face negotiation methods are used, then human agents can understand and plan negotiation strategies, but the process consumes excessive time and resources
Solution Approach 1:
The negotiation system enables agents to autonomously negotiate contract terms without human intervention. The automated agents independently analyze opponent behavior, generate negotiation strategies, and reach agreements, eliminating the need for human agents to manually understand and plan each negotiation interaction while maintaining negotiation quality through sophisticated AI algorithms
Solution Approach 2:
The patent replaces the mechanical human negotiation process with an automated electronic negotiation system. Human facial expressions, gestures, and verbal communications are substituted with digital agents that process negotiation data electronically, using algorithms to analyze opponent behavior and generate optimal negotiation moves, thereby dramatically reducing negotiation time while preserving decision quality
2Productivity
If reinforcement learning agents are trained with diverse behavioral patterns, then negotiation efficiency and scalability improve, but the system requires extensive training data from multiple domains
Solution Approach 1:
The patent segments the negotiation task into multiple independent behavioral models, each specialized in detecting and responding to specific opponent behavior patterns (e.g., selfish, prosocial, aggressive, cooperative). This segmentation allows the system to train smaller, more focused models on specific behavior types rather than requiring one massive model to handle all patterns, thereby reducing the overall training data requirement while maintaining high negotiation efficiency
Solution Approach 2:
The patent creates a universal negotiation agent that can adapt to multiple opponent behavior patterns through a set of interchangeable behavioral models. Instead of training separate specialized agents for each behavior type, the system develops a multi-functional agent capable of switching between different behavioral strategies based on the detected opponent type, reducing the need for extensive domain-specific training data for each scenario
3Device complexity
If a single behavioral model is used in negotiation agents, then the system is simpler to implement, but it reduces adaptability to different opponent negotiation styles
Solution Approach 1:
The patent implements a dynamic negotiation agent that can adapt its behavior in real-time based on the opponent's style. The system dynamically switches between different behavioral models (selfish, prosocial, aggressive, cooperative) depending on the detected opponent characteristics and negotiation context. This dynamic adaptation allows the agent to maintain simplicity in individual model structure while achieving high versatility through flexible model selection and combination
Solution Approach 2:
The patent creates a composite negotiation agent by combining multiple behavioral models into a unified system. Each behavioral model represents a different negotiation strategy, and the composite agent integrates these models to handle diverse opponent types. The system weighs and combines predictions from multiple behavioral models to generate optimal negotiation moves, achieving high adaptability while keeping individual component structures relatively simple
Data Source
AI summary
This disclosure relates generally to method and system for performing negotiation task using reinforcement learning agents. Performing negotiation on a task is a complex decision making process and to arrive at consensus on contents of a negotiation task is often expensive and time consuming due to the negotiation terms and the negotiation parties involved. The proposed technique trains reinforcement learning agents such as negotiating agent and an opposition agent. These agents are capable of performing the negotiation task on a plurality of clauses to agree on common terms between the agents involved. The system provides modelling of a selector agent on a plurality of behavioral models of a negotiating agent and the opposition agent to negotiate against each other and provides a reward signal based on the performance. This selector agent emulate human behavior provides scalability on selecting an optimal contract proposal during the performance of the negotiation task.