Entity Linking via Graph Distance and PageRank Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing entity linking systems face challenges such as ambiguity, scalability, and efficiency when processing large knowledge bases and documents, particularly in real-time, and struggle with limited information and evolving data.
Innovation Solution
The use of entity distance in a knowledge graph, employing Personalized PageRank techniques and vertex embeddings to measure distance between entities, facilitating accurate entity linking by optimizing a score function that considers local and global scores, and employing a heuristic optimization process for efficient candidate selection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional entity linking methods are used with large knowledge bases, then entity linking can be performed, but processing time increases and real-time performance deteriorates
Solution Approach 1:
The patent pre-computes and stores entity distances and rankings in the knowledge graph before actual entity linking operations. This preliminary computation of distance metrics and entity rankings enables fast query responses during real-time entity linking, avoiding the need to compute distances from scratch for each new query.
Solution Approach 2:
The patent replaces traditional mechanical search methods with personalized PageRank algorithms and vector embeddings to compute entity similarities. This substitution of computational approaches enables more efficient processing of large knowledge bases while maintaining accuracy in entity linking decisions.
2Productivity
If comprehensive entity linking is performed across large knowledge bases, then more entities can be linked, but system complexity increases
Solution Approach 1:
The patent computes and stores only the necessary entity distances and rankings locally relevant to each query context, rather than maintaining complex global computations across the entire knowledge graph. This localized approach enables comprehensive entity linking capability while reducing the computational complexity burden on the system.
3Productivity
If entity linking is performed with limited information and context, then processing speed increases, but accuracy deteriorates
Solution Approach 1:
The patent introduces entity distance as an intermediary metric that captures semantic relationships between entities without requiring extensive contextual information. This intermediary representation enables accurate entity linking even with limited input context by leveraging pre-computed distance relationships in the knowledge graph.
Data Source
AI summary
According to an embodiment, a method includes converting a knowledge base into a graph. In this embodiment, the knowledge base contains a plurality of entities and specifies a plurality of relationships among the plurality of entities, and entities in the knowledge base correspond to vertices in the graph, and relationships between entities in the knowledge base correspond to edges between vertices in the graph. The method may also include extracting a plurality of vertex embeddings from the graph. An example vertex embedding of the plurality of vertex embeddings represents, for a particular vertex, a proximity of the particular vertex to other vertices of the graph. Further, the method may include performing, based at least in part on the plurality of vertex embeddings, entity linking between input text and the knowledge base.