Focused Aggregation of Classification Model Outputs for Variable Length Documents
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital content management systems face challenges in accurately and efficiently classifying variable length documents, often resulting in false positives and inefficiencies due to rigid document length assumptions and resource-intensive training requirements.
Innovation Solution
The system employs a neural network-based classifier to determine category scores for document segments, focusing on positive trigger segments that meet a threshold score, and aggregates scores dynamically based on a positive trigger ratio to classify documents effectively, reducing false positives and resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional systems classify electronic documents using rigid document length assumptions, then the classification process is simplified, but the accuracy of classification deteriorates due to false positives
Solution Approach 1:
The patent divides electronic documents into multiple segments of varying lengths rather than treating documents as uniform fixed-length entities. Each segment is independently classified, allowing the system to handle variable length documents effectively while maintaining classification accuracy by focusing on relevant content portions.
2Adaptability or versatility
If conventional systems train classification models on large datasets, then model coverage improves, but resource consumption and training time increase significantly
Solution Approach 1:
The patent performs preliminary filtering by identifying positive trigger segments that meet threshold criteria before applying the full classification model. This preliminary action reduces the amount of data that requires intensive model training and processing, thereby reducing resource consumption while maintaining model coverage through focused training on relevant segments.
3Reliability
If conventional systems analyze all document segments equally, then comprehensive coverage is achieved, but processing time and computational resources increase
Solution Approach 1:
The patent applies different processing strategies to different segments based on their characteristics. Positive trigger segments that meet threshold criteria receive focused aggregation and weighted scoring, while other segments are processed differently or given less computational resources. This local quality approach maintains comprehensive coverage while improving processing efficiency by allocating resources strategically.
4Device complexity
If conventional systems use fixed aggregation methods for category scores, then the classification process is straightforward, but accuracy deteriorates due to false positives from irrelevant segments
Solution Approach 1:
The patent implements dynamic aggregation that adapts based on the proportion of positive trigger segments. When a high proportion of segments are positive triggers, the system applies focused aggregation with weighted scoring that emphasizes these segments. When the proportion is low, different aggregation strategies are used. This dynamic approach improves category score accuracy while maintaining reasonable process complexity through conditional logic.
Data Source
AI summary
Systems, methods, and non-transitory computer-readable media are disclosed for utilizing focused aggregation of classification model outputs to classify variable length documents. For instance, the disclosed systems can utilize a classification model to determine category scores for segments from an electronic document. Furthermore, the disclosed systems can identify positive trigger segments from the segments by comparing the category scores to a threshold category score. Moreover, the disclosed systems can determine a positive trigger ratio for the target category based on the positive trigger segments and the segments. Additionally, the disclosed systems can generate an aggregated category score for the electronic document from the positive trigger segments (when the positive trigger ratio satisfies a threshold positive trigger ratio) and distribute the electronic documents to client devices based on the aggregated category score.