Invoice Entity Extraction Override via User Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automatically processing digital invoices is challenging due to the difficulty in isolating and approving different invoice entities, and existing technologies lack effective methods for user feedback integration to correct machine learning model predictions.
Innovation Solution
A system that invokes an image processing module to ingest digital invoices, derives metrics, and uses a semantic document processing module with an entity extraction correction module to override machine learning model predictions based on user feedback from similar entities, producing a processed digital invoice for final disposition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If trained machine learning models are used for entity extraction, then processing automation is improved, but accuracy is worsened due to inability to adapt to user preferences
Solution Approach 1:
The system implements feedback loops where user corrections to entity extractions are captured and used to generate retraining data. This feedback mechanism allows the machine learning models to continuously improve accuracy based on actual user preferences and corrections, resolving the contradiction between automation and precision.
Solution Approach 2:
The system performs preliminary entity extraction using trained machine learning models to automate processing, then applies user feedback corrections to refine results. This preliminary action followed by correction allows the system to maintain high automation while improving accuracy through iterative refinement.
2Measurement precision
If user feedback is integrated to correct model predictions, then entity extraction accuracy is improved, but processing time is worsened
Solution Approach 1:
The system applies user feedback corrections selectively rather than to all extractions. By identifying and correcting only the most uncertain or error-prone entity extractions based on confidence thresholds and feedback patterns, the system improves accuracy while minimizing additional processing time.
3Measurement precision
If machine learning models are retrained with extensive data sampling, then entity extraction accuracy is improved, but device complexity is worsened
Solution Approach 1:
The system extracts only the relevant and most valuable data points from user feedback for model retraining, rather than processing all available data. By filtering and selecting only the most impactful corrections and examples, the system improves model accuracy while reducing the complexity of the retraining process.
Data Source
AI summary
A non-transitory computer readable storage medium has instructions executed by a processor to invoke an image processing module to ingest a digital invoice. An evaluation module derives metrics from the digital invoice. A semantic document processing module forms entity extracts from the digital invoice, where each entity extract from the digital invoice has a potential mapping to a trained machine learning model element. An entity extraction correction module overrides the potential mapping to the trained machine learning model element when user feedback from a similar entity extract from a previously processed digital invoice exists to produce a processed digital invoice with a user feedback element inconsistent with the potential mapping to the trained machine learning model element. The processed digital invoice is delivered to an accounting module for final disposition of the digital invoice.