Framework for emotion-enhanced language systems

Summary

The platform offers several significant advantages over existing approaches:

1. This invention represents the first approach to incorporate physiological, behavioral, cognitive, and psychological sensing from low-cost wearables to enhance the emotional understanding of LLMs. By integrating these multifaceted data sources, the platform provides a more comprehensive and nuanced understanding of human emotions.
2. The platform provides depth and intensity of emotions, addressing the limitations of existing sentiment analysis-based methods. This allows for a more accurate representation of complex emotional states, moving beyond simplistic positive/negative classifications.
3. By employing low-cost wearables to capture user emotional and cognitive states, the invention offers a cost-effective solution for emotional and cognitive assessment. This makes the technology accessible to a wider range of applications and users.
4. The platform offers precise transparency in identifying different emotional learnables that contribute to enhancing the emotional intelligence of LLMs. This clarity allows for better understanding and refinement of the emotional intelligence capabilities of the system.
5. The overall system facilitates the development of more empathetic and responsive chatbots, significantly improving human-LLM interactions. By incorporating a deeper understanding of user emotions and cognitive states, the platform enables LLMs to provide more contextually appropriate and emotionally intelligent responses.

These advantages collectively position the platform as a groundbreaking advancement in the field of emotional intelligence for LLMs, with the potential to revolutionize human-computer interactions across various domains.

Detailed Description

This invention discloses a platform designed to enhance the emotional intelligence of Large Language Models (LLMs) in chatbots. The platform addresses limitations in current approaches by incorporating multifaceted data from low-cost wearable devices to measure users’ cognitive, emotional, and behavioral states. The platform comprises two main stages:

Integration of these computed emotion scores with traditional inputs (user prompts, context, and chat history) to enhance LLM responses.

Quantification of emotion scores through low-cost wearable sensing of users’ cognitive, behavioral, and emotional states.

By leveraging physiological and behavioral data from wearable devices, the platform aims to provide a more comprehensive understanding of human emotions beyond simple sentiment analysis. This approach allows for a more nuanced interpretation of emotional states, including their intensity and complexity.

The invention utilizes correlations between cognitive load, biometric measurements, and detectable physiological and behavioral changes linked to specific emotions. It takes advantage of consumer-grade wearable health-tracking devices from companies like Fitbit, Garmin, Apple, Samsung, and Google to gather relevant data.

The platform represents a significant advancement in emotional intelligence for LLMS, enabling chatbots to respond more empathetically and effectively to users by understanding not just the textual content of interactions, but also the underlying human context. This framework has the potential to greatly improve human-computer interactions across various. applications and industries.