LLM Document Conversation Interface with Relevance Feedback

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current systems lack an effective method to enable users to interactively converse with a corpus of electronic documents using large language models, failing to efficiently extract and present relevant information in a user-friendly interface.

Innovation Solution

A system and method that utilize a large language model to process user queries, determine relevant documents, and present them in a user interface, allowing users to select documents, enter queries, and receive relevant information, with the system comprising hardware processors, client computing platforms, and components for document retrieval, relevance determination, and interface presentation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If users directly query a corpus of electronic documents using traditional search methods, then information can be retrieved, but the system cannot efficiently extract and present relevant information in a user-friendly interface

Engineering Contradiction:
Improveuser-friendly interfaceVSAvoidrelevant information extraction
Core Design Contradiction:
Ease of operationVSLoss of information

Solution Approach 1:

The patent introduces a large language model as an intermediary between the user's query and the document corpus. The LLM processes the query, identifies relevant documents, extracts key information, and presents it in a user-friendly format. This mediator resolves the contradiction by enabling both easy user interaction and efficient information extraction without requiring users to navigate complex search interfaces or manually analyze documents.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Loss of information

If the system processes multiple documents through a large language model, then relevant information can be extracted, but the computational complexity and processing time increase

Engineering Contradiction:
Improverelevant information extractionVSAvoidsystem complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent segments the document processing task by first identifying and selecting only the most relevant documents based on the query, then processing those specific documents through the large language model. This segmentation approach reduces the number of documents that require complex LLM processing, thereby lowering computational complexity and processing time while still extracting all necessary relevant information.

Inventive Principle:
Principle #1Segmentation

3Loss of information

If the system provides comprehensive information from multiple documents, then information completeness is improved, but the risk of hallucinations increases

Engineering Contradiction:
Improveinformation completenessVSAvoidhallucination risk
Core Design Contradiction:
Loss of informationVSReliability

Solution Approach 1:

The patent implements a feedback mechanism where the system monitors and tracks the sources of information provided to users. By maintaining provenance information that links extracted content back to specific documents and passages, the system can verify the authenticity of information and reduce hallucinations. This feedback loop ensures information completeness while maintaining reliability through source verification.

Inventive Principle:
Principle #23Feedback

4Ease of operation

If the system processes queries in real-time conversation mode, then user interaction is improved, but the processing speed and response time may be affected

Engineering Contradiction:
Improveinteractive conversationVSAvoidresponse time
Core Design Contradiction:
Ease of operationVSSpeed

Solution Approach 1:

The patent performs preliminary actions by pre-processing and indexing the document corpus before user queries are submitted. Documents are segmented, key features are extracted, and relevance criteria are established in advance. When users submit queries during conversation, the system can quickly match queries against pre-processed information, maintaining real-time interactive capabilities while ensuring fast response times.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentUS20240386037A1Systems and methods for providing user interfaces to converse with a corpus of electronic documents via a large language model
Publication Date: 2024.11.21 INSTABASE INC
  • US20240386037A1 patent drawing
  • US20240386037A1 patent drawing
  • US20240386037A1 patent drawing

AI summary

Systems and methods for providing user interfaces to converse with a corpus of electronic documents via a large language model are disclosed. Exemplary implementations may: present a user interface configured to obtain entry of user input from a user to select one or more documents to be provided as input to a large language model for an individual conversation; responsive to selection of the individual conversation, provide an individual query as a prompt to the large language model; obtain and present an individual reply from the large language model; determine an individual document from the one or more documents that is relevant to the individual reply; present the individual document in a particular portion of the user interface; and/or perform other steps.