Entity-Guided Prompting for Vertical-Domain Text Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large language models lack the ability to connect the intension and extension of domain-specific concepts, such as legal or medical, and establish relationships between different concepts, requiring additional input management for efficient vertical-domain language analysis.
Innovation Solution
A data processing method that involves obtaining texts, determining prompts based on entities within those texts, and using a language model to guide the recognition and determination of entities for improved analysis efficiency and precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If foundation models are directly applied to professional vertical domains, then general semantic understanding abilities are achieved, but the ability to connect intension and extension of domain-specific concepts is lacking
Solution Approach 1:
The patent segments the language analysis task into multiple independent NLP tasks (entity recognition, relationship extraction, attribute extraction) that can be processed separately using specialized models for each vertical domain, allowing precise connection of domain-specific concepts while maintaining general applicability
Solution Approach 2:
The patent introduces domain knowledge graphs as an intermediary structure that connects intension (concept definitions) and extension (specific instances) of domain-specific concepts, enabling foundation models to accurately understand and reason about vertical domain relationships without requiring direct retraining
2Measurement precision
If different NLP capabilities are invoked for processing distinct vertical domain concepts, then domain-specific accuracy is improved, but additional input management is required
Solution Approach 1:
The patent creates a unified input interface that accepts various vertical domain concepts and automatically routes them to appropriate NLP capabilities through the domain knowledge graph, allowing a single system to handle multiple domain-specific tasks without requiring separate input management processes for each domain
Data Source
Figure 1A
Figure 1B~1C
Figure 1D~2
AI summary
A data processing method may be applied to the field of artificial intelligence. The method includes: obtaining a first text and a second text, where the first text is a text that needs to be determined, and the second text is content for determining the first text; determining a first prompt prompt and a second prompt based on an entity included in the second text, where the first prompt indicates to recognize description of each entity in the first text, and the second prompt indicates to determine, as indicated by the second text, the first text based on the description of each entity; and obtaining, based on the first text, the first prompt, and the second prompt by using a language model, a result of determining the first text. In this application, the first prompt is used to guide the language model to recognize explanation for an entity in the first text, the language model automatically decomposes a task of understanding an element into prompts of an NLP task for input, and the language model is guided by using the prompts for text determining by using the description of the entity, to improve language analysis efficiency and precision.