Entity Linking in Incomplete Knowledge Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional entity linking approaches assume a complete knowledge graph and fail to distinguish between acronyms and non-acronyms, leading to inaccurate linking in incomplete or domain-specific knowledge graphs, particularly in enterprise contexts where acronyms may have different meanings.
Innovation Solution
A computing system computes relevance scores for entries in a knowledge graph based on identifiers and additional text context, using threshold scores specific to acronyms and non-acronyms to determine accurate links, and accounts for not-in-list entities, enabling domain-agnostic operation on private or enterprise knowledge graphs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional entity linking approaches are used, then the system operates on complete knowledge graphs, but the linking accuracy deteriorates when applied to incomplete or domain-specific knowledge graphs
Solution Approach 1:
The system adapts the entity linking approach based on the local characteristics of the knowledge graph. It detects whether the knowledge graph is complete or incomplete and applies different linking strategies accordingly, treating domain-specific graphs with specialized handling for acronyms and not-in-list entities
Solution Approach 2:
The system dynamically adjusts its entity linking behavior based on the type of identifier encountered. It distinguishes between acronyms and non-acronyms, applying different threshold scores and linking rules depending on the identifier category, making the system flexible rather than static
2Measurement precision
If conventional entity linking approaches are used, then the process is simple, but the precision of entity identification deteriorates due to failure to distinguish acronyms from non-acronyms
Solution Approach 1:
The system segments the entity linking process into distinct pathways based on identifier type. It separates acronym handling from non-acronym handling, with dedicated processing logic, threshold scores, and not-in-list entity detection rules for each category, thereby improving precision through structured differentiation
Solution Approach 2:
The system changes key parameters of the linking process based on identifier type. It adjusts threshold scores dynamically - using stricter thresholds for acronyms and different thresholds for non-acronyms - and modifies the not-in-list entity detection behavior accordingly, optimizing precision for each category
3Reliability
If the system handles not-in-list entities, then the accuracy improves, but the computational complexity increases
Solution Approach 1:
The system performs preliminary detection of not-in-list entities before finalizing entity links. It checks whether candidate entities exist in the knowledge graph and identifies not-in-list cases in advance, preventing incorrect linking and improving overall reliability through proactive validation
Data Source
AI summary
A computing system obtains computer-readable text comprising an identifier for an entity and an additional word. The computing system computes relevance scores for entries for entities in a computer-implemented knowledge graph upon the text, titles for the entities in the knowledge graph, and descriptions of the entities in the knowledge graph. The computing system determines whether the identifier for the entity is an acronym or a non-acronym. The computing system identifies an entry for the entity in the knowledge graph based upon the relevance scores and whether the identifier for the entity is an acronym or a non-acronym. The computing system generates a link between the identifier for the entity in the text and the entry for the entity in the knowledge graph, where data from the entry for the entity in the knowledge graph is presented to a user upon the link being selected.