US 12,450,274 B1
Adaptive information retrieval for multimodal data
Siddharth Jain, Mountain View, CA (US); Sivashanker Thiruchittampalam, Toronto (CA); Venkat Narayan Vedam, Mountain View, CA (US); and Jonathan Lin, Toronto (CA)
Assigned to INTUIT INC., Mountain View, CA (US)
Filed by INTUIT INC., Mountain View, CA (US)
Filed on Oct. 10, 2024, as Appl. No. 18/912,520.
Int. Cl. G06F 16/334 (2025.01); G06F 16/35 (2025.01)
CPC G06F 16/3347 (2019.01) [G06F 16/35 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
receiving, by at least one processor, a document including a plurality of sections;
generating, by the at least one processor, a classification of the document according to at least one of:
a structure of at least one of the plurality of sections,
a hierarchy of the plurality of sections, and
a content of at least one of the plurality of sections;
determining, by the at least one processor, a chunking strategy optimized for the classification from among a plurality of available chunking strategies;
dividing, by the at least one processor, the document into a plurality of chunks according to the chunking strategy;
generating, by the at least one processor, respective vector representations of respective ones of the plurality of chunks; and
provisioning, by the at least one processor, at least one database comprising the vector representations, wherein the database is configured to supply the vector representations to retrieval augmented generation (RAG) operations performed by at least one large language model (LLM).