Abstract
When artificial intelligence (AI) agents search large product catalogs, conventional retrieval systems often return the wrong document because they compress each document into a single summary vector, losing small but critical details that distinguish one product from another. Proposed herein is a spectral retriever technique that addresses this problem by examining documents at multiple levels of detail at the same time, from individual product identifiers to overall document meaning. The spectral retriever technique automatically determines whether a query calls for fine-grained matching, broader semantic understanding, or an intermediate level of granularity, without domain-specific rules or model retraining.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Morandi, Andrea and Gupta, Saurabh, "SPECTRAL RETRIEVER FOR LARGE LANGUAGE MODEL (LLM) MULTI AGENT SYSTEMS", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10212