Hierarchical String Matching Framework for Database Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current database mapping prediction techniques face challenges in efficiently and accurately matching input strings to ground-truth database tables, often requiring extensive computational resources and sacrificing predictive accuracy for speed.
Innovation Solution
A hierarchical string matching machine learning framework is introduced, combining table-independent and table-dependent match models, including exact, probabilistic, disjoint, and embedding-based models, to generate database mapping predictions. This framework limits the use of computationally expensive machine learning models to instances where rule-based operations fail, improving computational efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If computationally expensive machine learning models are used for database mapping prediction, then predictive accuracy is improved, but computational efficiency deteriorates
Solution Approach 1:
The patent segments the matching process into multiple hierarchical levels: exact match models first, then probabilistic match models, followed by disjoint match models, and finally embedding-based match models. This segmentation allows the system to use simpler models for common cases and reserve computationally expensive models for difficult cases, resolving the contradiction between accuracy and efficiency.
Solution Approach 2:
The patent applies preliminary filtering using exact match models and probabilistic match models before invoking computationally expensive embedding-based match models. This preliminary action eliminates many cases that don't require complex processing, improving overall computational efficiency while maintaining accuracy for cases that do need sophisticated matching.
2Productivity
If rule-based operations are used for string matching, then computational efficiency is improved, but predictive accuracy deteriorates
Solution Approach 1:
The patent merges rule-based matching operations with machine learning-based matching operations in a hybrid framework. Exact match models and probabilistic match models provide rule-based efficiency, while embedding-based match models provide machine learning-based accuracy. The combination resolves the contradiction by leveraging the strengths of both approaches.
Solution Approach 2:
The patent introduces intermediate matching models (exact match and probabilistic match models) that act as mediators between simple rule-based operations and complex machine learning models. These intermediaries handle cases that are too complex for simple rules but don't require the full power of expensive machine learning models, resolving the accuracy-efficiency tradeoff.
3Measurement precision
If extensive computational resources are allocated to database mapping prediction, then predictive accuracy is improved, but resource usage deteriorates
Solution Approach 1:
The patent applies partial action by using computationally expensive embedding-based match models only when necessary, rather than applying them to all cases. The hierarchical framework determines the appropriate level of processing for each input string, allocating extensive computational resources only when simpler models fail to provide sufficient accuracy, thus resolving the contradiction between accuracy and resource usage.
4Productivity
If simple matching algorithms are used, then computational efficiency is improved, but predictive accuracy deteriorates
Solution Approach 1:
The patent implements a dynamic matching system that adapts the complexity of algorithms used based on the specific input characteristics. The hierarchical framework dynamically selects from exact match, probabilistic match, disjoint match, or embedding-based models depending on what is appropriate for each case, resolving the static contradiction between simple algorithm efficiency and complex algorithm accuracy.
Data Source
AI summary
Various embodiments of the present invention provide methods, apparatus, systems, computing devices, computing entities, and/or the like for performing predictive mapping operations with respect to a ground-truth database table. Certain embodiments of the present invention utilize systems, methods, and computer program products that perform predictive mapping operations utilizing a hierarchical string matching machine learning framework using at least one or more of an exact match model, a probabilistic match model, a disjoint match model, and an embedding-based match model.