Cascade Geocoding Scoring Engine for Address Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing geocoding systems face challenges in accurately matching input addresses due to the incompleteness of point level data sets, leading to false positive matches when combined with centerline data sets, as they often assume point level data is complete, resulting in reduced match accuracy and increased false positives.
Innovation Solution
A cascade geocoding scoring and matching engine processes both point level and centerline data sets to identify candidate matches, assigning scores and penalties to determine the best match, prioritizing exact matches in point level data and combining scores from both sets to minimize false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If only point level data set is used for geocoding, then geocoding speed is improved, but match accuracy deteriorates due to incomplete data causing false positives
Solution Approach 1:
The geocoding process is segmented into distinct phases: first querying the point level data set for rapid matching, then conditionally querying the centerline data set only when needed. This segmentation allows the system to achieve high speed for complete addresses while maintaining accuracy by falling back to centerline interpolation when point level data is insufficient or causes false positives.
Solution Approach 2:
The system performs preliminary action by first checking the point level data set before consulting the centerline data set. This preliminary query filters out most cases quickly, and only proceeds to the more computationally intensive centerline processing when necessary, thereby maintaining both speed and accuracy.
2Reliability
If both point level and centerline data sets are processed separately by independent geocoding engines, then completeness of address matching is improved, but false positive matches increase
Solution Approach 1:
The system merges the results from point level and centerline data set queries into a unified candidate list. By combining both data sources and their respective matches, the system ensures completeness of address matching while maintaining the ability to evaluate all candidates together to avoid false positives.
Solution Approach 2:
The system implements feedback by using the point level match results to inform the centerline query process. When a point level match is found, the system uses that information to constrain or guide the centerline matching process, thereby reducing false positives while maintaining completeness.
3Ease of operation
If point level data is assumed to be complete, then processing simplicity is improved, but match accuracy deteriorates due to false positives from incomplete data
Solution Approach 1:
The system dynamically changes the completeness parameter assumption based on the query results. Instead of assuming point level data is always complete or never complete, the system adjusts its behavior based on whether matches are found and whether multiple candidates exist, thereby maintaining simplicity while improving accuracy.
Data Source
AI summary
A method for processing two different types of geographic address data sets to establish candidate matched data includes entering address data into the system and accessing a point level address data set. A determination is made if the point level data set contains an exact match to the entered address data. Any exact address data match from the point level data set is output. Candidate address data matches from the point level data set are saved when no exact match is determined as existing in the point level data set. A centerline address data set is accessed and a determination is made of candidate address data matches from the centerline address data set. Candidate address data from both the point level data set and the centerline data set are scored and matched.A system for geocoding and address hygiene includes a processing system with an input and an output system coupled to said processing system. A point level data set is coupled to the processing system. A centerline data set is coupled to the processing system. A cascade geocoding scoring and matching engine is coupled to the processing system.