Entity Disambiguation via Probabilistic Attribute Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the era of large electronic environments, identifying whether online mentions or database profiles belong to a specific individual with a unique name is challenging due to the lack of personalized tools that can distinguish between individuals with similar names, leading to unclear information retrieval.
Innovation Solution
The use of statistical functions to make probabilistic deductions about entity attributes, constructing optimal combinations of these attributes to generate search queries that return more precise results by assigning reference and attribute scores, and employing a cohort-based framework to disambiguate entity reflections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional search methods are used to retrieve information about individuals, then search coverage is achieved, but precision is poor due to inability to distinguish between individuals with same or similar names
Solution Approach 1:
The patent segments the entity identification process into multiple stages: extracting entity mentions from text, computing entity scores based on multiple attributes (name, location, occupation, etc.), and progressively refining results through iterative searching. This segmentation allows the system to handle the complex task of disambiguation through manageable steps, improving precision without overwhelming system complexity
Solution Approach 2:
The patent introduces multiple dimensions for entity identification beyond simple name matching. It computes entity scores based on multiple attributes including location, occupation, age, and other contextual information. This multi-dimensional approach transforms the problem from one-dimensional name matching to multi-dimensional entity profiling, significantly improving search precision
2Measurement precision
If multiple entity attributes are collected and analyzed, then entity disambiguation accuracy improves, but processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary actions by pre-computing entity scores based on multiple attributes before actual search queries are executed. It extracts and analyzes entity mentions in advance, building a foundation of entity profiles that can be quickly queried later. This preliminary processing reduces the computational burden during actual search operations, balancing accuracy with processing time
Solution Approach 2:
The patent implements a progressive refinement approach where it initially processes a subset of entity attributes and progressively adds more attributes as needed. The system computes entity scores using available attributes and iteratively refines results, rather than processing all possible attributes simultaneously. This partial action approach reduces initial processing time while still achieving high accuracy through iterative improvement
Data Source
AI summary
The technology disclosed relates to identifying entity reflections that refer to a same real-world entity. In particular, it relates to using statistical functions to make probabilistic deductions about entity attributes, which are used to construct optimal combinations of entity attributes. These optimal combinations of entity attributes are further used to generate search queries that return more precise search results with greater recall.