Attribute Ranking in Structured Knowledge Bases

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Automating the creation of entity-specific pages in online encyclopedias that provide a succinct summary of reference topics is challenging, especially when dealing with millions of entities, as existing methods struggle to rank and select relevant attributes effectively.

Innovation Solution

The architecture computes intermediate features such as entity type importance, inverse document frequency, and attribute popularity, then uses a machine-learned classifier to generate relevance scores for attributes, allowing for efficient ranking and selection of key attributes based on entity type and human judgments.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If automated methods are used to create entity-specific pages for millions of entities, then productivity is improved, but the ability to accurately rank and select relevant attributes deteriorates

Engineering Contradiction:
Improveautomation scalabilityVSAvoidattribute relevance ranking accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent segments the attribute ranking problem into multiple independent features that can be computed separately and then combined. Each feature (entity type importance, attribute IDF, entity type IDF, attribute popularity) addresses a specific aspect of relevance, allowing the system to handle millions of entities efficiently while maintaining accuracy through the cumulative effect of multiple segmented features

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces intermediate features as mediators between the raw data and the final relevance score. These intermediate features (entity type importance, attribute IDF, etc.) serve as computational bridges that transform raw entity-attribute data into meaningful relevance signals, enabling scalable automated processing while preserving ranking accuracy

Inventive Principle:
Principle #24Intermediary (Mediator)

2Quantity of substance

If all available attributes are included in entity summaries, then completeness is improved, but information quality deteriorates due to inclusion of obscure and not useful attributes

Engineering Contradiction:
Improvenumber of attributesVSAvoidinformation quality
Core Design Contradiction:
Quantity of substanceVSLoss of information

Solution Approach 1:

The patent applies local quality by making the attribute selection process entity-specific rather than uniform. The relevance score for each attribute is computed individually based on entity-type-specific features, allowing the system to select different subsets of attributes for different entities. This ensures that each entity displays only the attributes that are locally relevant to it, filtering out obscure attributes while maintaining completeness for important ones

Inventive Principle:
Principle #3Local quality

3Measurement precision

If manual creation of topic summaries is used, then attribute relevance accuracy is improved, but productivity deteriorates

Engineering Contradiction:
Improveattribute relevance accuracyVSAvoidcreation speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent implements self-service by enabling the system to automatically compute relevance scores and select attributes without human intervention. The machine learning model uses entity-type-specific features to autonomously determine which attributes are relevant for each entity, allowing rapid processing of millions of entities while maintaining accuracy comparable to manual curation through the quality of its feature-based scoring mechanism

Inventive Principle:
Principle #25Self-service

Data Source

PatentUS9229988B2Ranking relevant attributes of entity in structured knowledge base
Publication Date: 2016.01.05 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

Architecture that addresses at least the problem of ranking the relevant attributes for a given entity within the context of a structured knowledge base (SKB). The architecture utilizes the attribute, entity type statistics, and the taxonomy of the attributes to consistently and efficiently rank attributes for each and every type of entity in the SKB. Using the SKB, intermediate features are computed, including the importance or popularity each entity type for every entity, inverse document frequency (IDF) computation for each attribute on a global basis, IDF computation for entity types, and the popularity of attributes for each entity type. The intermediate features are aggregated to obtain a final feature set, which can be used in combination with human judgments to train a machine learned classifier model to produce and predict a relevance score for a given entity and each of its attributes. The attributes are ranked for each entity using this score.