Entity identification and entity attribute verification using large language models
Large language models with validated responses enhance the efficiency and accuracy of entity identification and attribute verification by automating data source analysis, reducing manual errors and improving scalability.
CA3299772A1Pending Publication Date: 2026-09-21EQUIFAX INC
0 Cites 0 Cited by
Patent Information
- Application Number
- CA3299772
- Authority / Receiving Office
- CA · CA
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-01-28
- Filing Date
- 2026-01-26
- Publication Date
- 2026-09-21
AI Technical Summary
Technical Problem
Existing systems rely on manual examination of data sources for entity identification and attribute verification, leading to inefficiencies, inaccuracies, and scalability limitations.
Method used
Employing large language models to identify entities and verify entity attributes by using prompts with false data items to validate responses, ensuring accuracy and reducing manual intervention.
Benefits of technology
Improves scalability and reduces errors by automating the process of entity identification and attribute verification, ensuring the accuracy of information used in decision-making systems.
✦ Generated by Eureka AI based on patent content.
Abstract
A system can identify entities and verify entity attributes by leveraging the ability of trained large language model (LLM) to extract entity information from data instances of one or more data sources. The data instances may be, for example, writings such as news articles. The writings may include information about entities and / or events in which the entities were involved. The system may generate a pair of input prompts for the LLM to cause the LLM to process a data instance(s). Each prompt may include a deliberately false data item and the LLM may be caused to generate a response for each prompt. The validity of the information in the responses can be verified by determining that false data items in both responses match. The system can configure another computing system to utilize information about a given entity identified in the data instance and included in the responses.
Need to check novelty before this filing date? Find Prior Art