Salient Entity Tagging via Reference Position Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing technologies fail to accurately identify and highlight salient entities within articles, leading to inefficiencies in content presentation and user engagement.

Innovation Solution

The method involves analyzing articles to identify entity terms, determining relevance scores, and generating salient entity tags based on reference positions, allowing for the selection and display of key entities within the content.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If existing technologies are used to identify entities in articles, then the process can be performed, but the accuracy and precision of identifying salient entities is insufficient

Engineering Contradiction:

Improveaccuracy of identifying salient entitiesVSAvoidreliability of entity identification

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The patent segments the entity identification process into multiple stages: first identifying all entity terms in the article, then calculating relevance scores for each entity term, and finally selecting salient entities based on both relevance scores and reference position information. This segmentation allows each stage to focus on specific aspects, improving overall accuracy and precision.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by considering the reference position information of each entity term within the article structure. Different entity terms are evaluated based on their specific positions and contexts in the article, allowing the system to identify salient entities that are locally significant rather than uniformly important throughout the text.

Inventive Principle:

Principle #3Local quality

2Measurement precision

If multiple entity terms are identified and evaluated, then the precision of salient entity identification improves, but the complexity of the process increases

Engineering Contradiction:

Improveprecision of salient entity identificationVSAvoidcomplexity of entity analysis process

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary actions by first extracting all entity terms from the article and pre-calculating their relevance scores before final selection. This preliminary processing organizes the data in advance, making the final selection of salient entities more efficient and less complex, while still maintaining high precision through the multi-criteria evaluation.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11803605B2Determining salient entities and generating salient entity tags based upon articles

Publication Date: 2023.10.31 YAHOO AD TECH LLC

AI summary

In an example, an article may be analyzed to identify entity terms. Entity term relevance scores associated with the entity terms may be determined based upon the article and the entity terms. One or more first entity terms may be selected based upon the entity term relevance scores. One or more sets of reference position information associated with the one or more first entity terms may be determined. A first set of reference position information is based upon one or more positions, in the article, of one or more references to a first entity term. One or more second entity terms of the one or more first entity terms may be selected based upon the one or more sets of reference position information. A set of one or more salient entity tags associated with the article may be generated based upon the one or more second entity terms.