Saved in:
| Main Author: | |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2410.10039 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929540764270592 |
|---|---|
| author | Rasal, Sumedh |
| author_facet | Rasal, Sumedh |
| contents | In recent years, large language models have demonstrated remarkable capabilities in natural language understanding and generation. However, these models often struggle with hallucinations and maintaining long term contextual relevance, particularly when dealing with private or local data. This paper presents a novel architecture that addresses these challenges by integrating an orchestration engine that utilizes multiple LLMs in conjunction with a temporal graph database and a vector database. The proposed system captures user interactions, builds a graph representation of conversations, and stores nodes and edges that map associations between key concepts, entities, and behaviors over time. This graph based structure allows the system to develop an evolving understanding of the user preferences, providing personalized and contextually relevant answers. In addition to this, a vector database encodes private data to supply detailed information when needed, allowing the LLM to access and synthesize complex responses. To further enhance reliability, the orchestration engine coordinates multiple LLMs to generate comprehensive answers and iteratively reflect on their accuracy. The result is an adaptive, privacy centric AI assistant capable of offering deeper, more relevant interactions while minimizing the risk of hallucinations. This paper outlines the architecture, methodology, and potential applications of this system, contributing a new direction in personalized, context aware AI assistance. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_10039 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | A Multi-LLM Orchestration Engine for Personalized, Context-Rich Assistance Rasal, Sumedh Multiagent Systems In recent years, large language models have demonstrated remarkable capabilities in natural language understanding and generation. However, these models often struggle with hallucinations and maintaining long term contextual relevance, particularly when dealing with private or local data. This paper presents a novel architecture that addresses these challenges by integrating an orchestration engine that utilizes multiple LLMs in conjunction with a temporal graph database and a vector database. The proposed system captures user interactions, builds a graph representation of conversations, and stores nodes and edges that map associations between key concepts, entities, and behaviors over time. This graph based structure allows the system to develop an evolving understanding of the user preferences, providing personalized and contextually relevant answers. In addition to this, a vector database encodes private data to supply detailed information when needed, allowing the LLM to access and synthesize complex responses. To further enhance reliability, the orchestration engine coordinates multiple LLMs to generate comprehensive answers and iteratively reflect on their accuracy. The result is an adaptive, privacy centric AI assistant capable of offering deeper, more relevant interactions while minimizing the risk of hallucinations. This paper outlines the architecture, methodology, and potential applications of this system, contributing a new direction in personalized, context aware AI assistance. |
| title | A Multi-LLM Orchestration Engine for Personalized, Context-Rich Assistance |
| topic | Multiagent Systems |
| url | https://arxiv.org/abs/2410.10039 |