Semantic Data Joining via Statistical Co-occurrence

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional join processing in databases is inefficient and inconvenient, especially when data sets have different representations, as it requires manual intervention to find intermediate mapping tables, and existing systems struggle to perform joins automatically in such scenarios.

Innovation Solution

The system predicts semantic relationships between data sets using statistical co-occurrence scores based on historical data, enabling automated semantic joins without the need for intermediate mapping tables by identifying pairs of values and calculating aggregate correlation scores to infer join relationships.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional join processing is used with string equality comparisons, then join accuracy is maintained for curated data, but manual intervention is required to find intermediate mapping tables and the process becomes time-consuming

Engineering Contradiction:
Improvejoin accuracyVSAvoidtime to find mapping tables
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs self-service by automatically computing statistical co-occurrence scores between values in different data sets and inferring semantic relationships without requiring manual intervention to find mapping tables. The processor autonomously identifies pairs of values, calculates their co-occurrence statistics, and determines semantic relationships, eliminating the time-consuming manual process while maintaining join accuracy through statistical analysis.

Inventive Principle:
Principle #25Self-service

2Adaptability or versatility

If manual join processing is performed for data sets with different representations, then semantic relationships can be established, but the process becomes complex and infeasible under some circumstances

Engineering Contradiction:
Improveability to join different representationsVSAvoidcomplexity of manual join process
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent replaces the mechanical manual process of finding and applying mapping tables with an automated statistical computing system. The processor automatically computes statistical co-occurrence scores for pairs of values from different data sets and infers semantic relationships, substituting the complex manual mechanical process with an automated computational approach that handles diverse representations systematically and efficiently.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If intermediate mapping tables are used for semantic joins, then joins between different representations are enabled, but the requirement for multiple mapping tables increases system complexity

Engineering Contradiction:
Improvesemantic join capabilityVSAvoidnumber of mapping tables
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent extracts the essential semantic relationship information directly from the data sets by computing statistical co-occurrence scores between pairs of values, eliminating the need for separate intermediate mapping tables. Instead of requiring multiple mapping tables to bridge different representations, the system directly analyzes the statistical relationships within the data itself, extracting the necessary join information without auxiliary structures.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP3304347B1Joining semantically-related data using big table corpora
Publication Date: 2019.09.04 MICROSOFT TECHNOLOGY LICENSING LLC
  • EP3304347B1 patent drawingFigure 1
  • EP3304347B1 patent drawingFigure 2
  • EP3304347B1 patent drawingFigure 3

AI summary

Examples of the disclosure enable performing semantic joins using a big table corpus. Pairs of values from at least two data sets are identified. The pairs of values include one value from a first one of the data sets and one value from a second one of the data sets. Statistical co-occurrence scores for the identified pairs of values are determined based on historical co-occurrence data. The determined statistical co-occurrence scores are used for predicting a semantic relationship between the at least two data sets. The predicted semantic relationship is used for joining the at least two data sets.