Method and device for realizing knowledge graph concept recognition from the perspective of mining multi-granularity concept features
By mining concept features from a multi-grained perspective in the knowledge graph, the problem of the lack of cognitive ability of existing natural language understanding methods is solved, and the machine's ability to understand and distinguish concepts is realized, and the effect of natural language understanding is improved.
CN118820483BActive Publication Date: 2025-03-28YANGTZE NORMAL UNIVERSITY
View PDF 0 Cites 0 Cited by
Patent Information
- Application Number
- CN202410784643.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-18
- Publication Date
- 2025-03-28
- Estimated Expiration
- 2044-06-18
AI Technical Summary
Technical Problem
The existing natural language understanding methods are based on information retrieval and matching, and do not have the same cognitive ability as humans, and cannot effectively understand and distinguish concepts.
Method used
By mining conceptual features in the knowledge graph from a multi-grained perspective, including the presence/absence of attributes, attribute value types and attribute values, the frequent attributes and attribute values of concepts are generated, and the human cognitive process of concepts is simulated.
Benefits of technology
The machine's ability to understand and distinguish concepts is realized, so that machines can have concept cognitive abilities like humans and improve the effect of natural language understanding.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN118820483B_ABST
Abstract
The present invention relates to a method and device for realizing knowledge graph concept cognition from the perspective of mining multi-granularity concept features, and belongs to the field of artificial intelligence technology. The method includes: mining frequent attribute features of an attribute information system and generating a maximum frequent attribute pattern; mining the maximum frequent attribute pattern and the corresponding attribute value feature of an attribute type or attribute value information system based on the maximum frequent attribute pattern of an attribute or attribute type information system; if the attribute value feature corresponding to the current maximum attribute pattern is frequent, then end its task in the attribute type or attribute value information system; if the attribute value feature corresponding to the current maximum attribute pattern is not frequent, then divide it into multiple sub-patterns, and then only the sub-patterns that are not repeated and not included in the candidate set will participate in subsequent judgments; derive all frequent attributes except itself through the maximum attribute pattern, and generate corresponding frequent attribute value features. The present invention can simulate human cognitive ability of concepts.
Need to check novelty before this filing date? Find Prior Art