Abstract
Current large language model (LLM)-based platforms offer interactive user experiences with limited personalization. Although these LLM-based platforms can leverage publicly available and trained data to provide contextual responses to user queries, they cannot securely leverage private data about the user to personalize the experience. As a result, current systems lack the ability to provide context-sensitive and personalized content that meets specific user requirements or preferences. Moreover, robust privacy safeguards are needed to properly leverage and optimize a user’s private data for content delivery. To address these issues, proposed herein are techniques for providing an application programming interface (API) service that interfaces with a vector database (vector DB) to obtain private user profile data and generate a secure, portable, and personalized encoded user profile that can be used by LLM-based platforms to optimize and deliver context-aware responses to user queries.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Lui, Cameron; Ibraimova, Aizhan; White, David C.; Survase, Rajeshri Prashant; Chihani, Bachir; and Pham, Tiffany K., "OPTIMIZING LLM-BASED QUERIES AND RESPONSES USING SECURE, PORTABLE, AND PERSONALIZED ENCODED USER PROFILE", Technical Disclosure Commons, (May 06, 2024)
https://www.tdcommons.org/dpubs_series/6967