Weighted Feature Vectors for Genealogical Identity Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Genealogical databases face challenges in identifying duplicate individuals due to discrepancies in historical records and human accounts, leading to difficulties in aligning ancestor data accurately.
Innovation Solution
A machine-learning based system that extracts features from tree persons, calculates metrics, and generates a weighted score to determine if two individuals correspond to the same real-life individual, using a training process to refine feature weights for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional manual methods are used to identify duplicate individuals in genealogical databases, then data accuracy may be maintained through human review, but the process becomes time-consuming and inefficient when dealing with large volumes of historical records
Solution Approach 1:
The patent replaces manual human review processes with an automated machine learning system that uses feature extraction, metric calculation, and weighted scoring to identify duplicate individuals. The system extracts features from historical records, calculates similarity metrics, and generates scores to automatically determine duplicates, eliminating the need for time-consuming manual comparison while maintaining identification accuracy through algorithmic precision
Solution Approach 2:
The system creates structured feature representations (copies) of unstructured historical record data, transforming diverse historical documents into standardized feature vectors that can be systematically compared. This copying process converts various formats of historical records into a uniform structure that enables efficient automated comparison without losing essential information
2Quantity of substance
If multiple historical records with discrepancies are aggregated, then more comprehensive data is obtained, but the reliability of individual person data decreases due to conflicting information
Solution Approach 1:
The patent applies different weighting strategies to different features based on their reliability and importance. Rather than treating all data points equally, the system assigns higher weights to more reliable features (such as exact match fields like birth dates) and lower weights to less reliable features (such as name variations). This local quality approach allows comprehensive data aggregation while maintaining overall data consistency through differential weighting
Solution Approach 2:
The system dynamically adjusts feature weights based on the specific comparison context and training data. The weighted sum calculation changes parameters (weights) according to the reliability of each feature type, allowing the system to adapt to different data quality scenarios. This parameter change mechanism enables the system to handle conflicting information by adjusting the influence of each data source based on its perceived reliability
3Speed
If a simple matching algorithm is used, then processing speed is maintained, but the ability to handle complex discrepancies in historical records is insufficient
Solution Approach 1:
The patent segments the duplicate identification process into distinct modular components: feature extraction, metric calculation, and weighted score generation. Each segment handles specific aspects of the comparison independently, allowing the system to process complex discrepancies through a series of simple, fast operations rather than requiring a single complex algorithm. This segmentation maintains processing speed while improving adaptability
Solution Approach 2:
The system transforms the duplicate identification problem from a single-dimension string matching task into a multi-dimensional feature space comparison. By extracting multiple features (names, dates, locations, relationships) and calculating metrics across these dimensions with different weights, the system gains the ability to handle complex discrepancies while maintaining processing efficiency through vector-based operations
4Device complexity
If feature weights are manually assigned, then the system remains simple and easy to configure, but accuracy decreases when facing diverse genealogical data patterns
Solution Approach 1:
The patent implements a feedback mechanism where the system learns from training data to automatically optimize feature weights. The machine learning model analyzes labeled examples of duplicate and non-duplicate pairs, receives feedback on prediction accuracy, and adjusts weights accordingly. This feedback loop enables the system to improve accuracy for diverse genealogical data patterns without increasing operational complexity, as the weight optimization occurs during the training phase rather than during deployment
Data Source
AI summary
Systems and methods for determining whether two tree persons in a genealogical database correspond to the same real-life individual. Embodiments include identifying two tree persons in a genealogical database and extracting a plurality of features from both tree persons to generate two vectors. Embodiments also include calculating a plurality of metrics between the two vectors to generate a metric function. Embodiments further include generating feature weights using a recursive process based on training data input by external users, and generating a score by calculating a weighted sum of the metric function being weighted by the feature weights. The generated score may then be compared to a threshold value.