Semantic Refinement Mechanism for Asset Classification Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual classification of assets in business glossaries often results in inaccurate and inconsistent classifications due to 'first best fit' approaches, and existing automatic classification methods rely on problematic manual assignments, leading to inefficiencies in merging and reconciling assets across organizations.
Innovation Solution
A computer-implemented method using a semantic refinement mechanism to evaluate and refine asset classifications based on a business glossary's hierarchy and domain ontology, allowing for more specific classifications and a feedback-enabled weighted calibration mechanism to iteratively adjust and undo refinements as necessary.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual classification is used, then classification accuracy can be maintained through domain knowledge, but classification consistency and precision deteriorate due to 'first best fit' approaches
Solution Approach 1:
The system implements a feedback mechanism where automated classification results are evaluated against refinement criteria, and the outcomes feed back into improving future classifications. The refinement process uses feedback from analyzing similar assets and their classifications to iteratively improve classification precision while maintaining accuracy.
Solution Approach 2:
The patent replaces manual mechanical classification processes with an automated system that uses computer processors to evaluate refinement criteria, analyze asset attributes, and generate refined classifications. This substitution eliminates human bias in 'first best fit' approaches while maintaining domain knowledge through trained algorithms.
2Productivity
If automatic classification is used, then classification speed improves, but classification accuracy deteriorates due to reliance on problematic manual assignments for training
Solution Approach 1:
The system performs preliminary refinement of manual classifications before using them for training automated classifiers. By pre-processing and correcting manual assignments through the refinement mechanism, the system prepares high-quality training data that enables accurate automated classification without relying on inherently problematic manual labels.
Solution Approach 2:
The automated classification system uses feedback from refined classifications to continuously improve its performance. The system evaluates automated results against refinement criteria and adjusts its training models accordingly, creating a feedback loop that improves both speed and accuracy over time.
3Reliability
If manual classification is used, then domain expertise is leveraged, but classification consistency across assets deteriorates
Solution Approach 1:
The system applies local quality by evaluating each asset's classification against refinement criteria specific to that asset's attributes and context. Rather than applying a uniform classification rule, the system tailors the refinement process to each asset's specific characteristics, ensuring consistent and reliable classifications that adapt to local domain requirements.
Solution Approach 2:
The patent replaces manual classification consistency with an automated evaluation mechanism that objectively assesses each classification against defined refinement criteria. This substitution ensures consistent application of classification standards across all assets while maintaining domain expertise through the structured refinement process.
4Ease of manufacture
If existing manual classifications are used for training, then training data is readily available, but training quality deteriorates due to naive assignments
Solution Approach 1:
The system performs preliminary refinement of manual classifications to create high-quality training data before it is used for model training. This pre-processing step ensures that the training data reflects accurate and consistent classifications rather than naive assignments, while still utilizing the readily available manual classification data.
Solution Approach 2:
The training process incorporates feedback from refined classifications to continuously improve training data quality. The system evaluates and corrects manual assignments, using this feedback to generate higher-quality training datasets that enable more accurate automated classification models.
Data Source
AI summary
Techniques are described for refining the manual classification of assets classified or categorized using the terms of a business glossary. A semantic refinement mechanism is used to refine the manual classification of such assets, as well as subsequently evaluate the refined asset classifications. Further, the refined asset classifications may be used as a training set for a machine learning classifier. That is, should the classification of an asset contributing to a refinement change, the refinement based on that classification may be undone, at least in some cases.