Hierarchical Attention Keyword Classifier for Document Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current document classification methods face inefficiencies and reliability issues in performing context-based predictions, particularly in healthcare-related applications, due to high computational loads and false positives from urgent/expediated keywords.
Innovation Solution
The implementation of a hierarchical attention-based keyword classifier machine learning framework that generates contextual keyword classifications using an attention-based encoder and classifier model, delaying document classification until affirmative contextual keyword classifications are determined, and utilizing exclusionary page templates to avoid false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional document classification methods are used, then all documents are classified in real-time, but computational load increases and false positives occur from urgent/expediated keywords
Solution Approach 1:
The patent segments the document classification process into two distinct phases: (1) a keyword screening phase that identifies documents containing urgent/expediated keywords, and (2) a full classification phase that applies only to documents passing the keyword screen. This segmentation reduces computational load by avoiding full classification on all documents while maintaining reliability through targeted analysis of suspicious cases.
Solution Approach 2:
The patent implements preliminary keyword-based filtering before full document classification. By performing this preliminary action on urgent/expediated keywords, the system identifies and flags specific documents for further classification, thereby reducing the overall computational burden while ensuring that potentially important documents are not missed.
2Productivity
If full document classification is performed on all documents, then classification coverage is complete, but processing time and computational resources increase
Solution Approach 1:
The patent applies partial classification action by performing keyword-based screening on all documents (excessive action for filtering) and full classification only on a subset of documents that match urgent/expediated keywords (partial action for detailed analysis). This approach optimizes processing throughput by avoiding redundant full classification while ensuring timely identification of critical documents.
3Measurement precision
If urgent/expediated keywords are used for classification, then important documents are identified, but false positives increase
Solution Approach 1:
The patent introduces an intermediary verification step between keyword matching and final classification. Documents containing urgent/expediated keywords are not immediately classified as high-priority but are instead flagged for further review or contextual analysis. This intermediary action reduces false positives by verifying the actual relevance of keyword matches before making classification decisions.
Data Source
AI summary
Various embodiments of the present invention provide methods, apparatus, systems, computing devices, computing entities, and/or the like for performing context-based document classification prediction using a hierarchical attention-based keyword classifier machine learning framework. Certain embodiments of the present invention utilize systems, methods, and computer program products that perform context-based document classification prediction using at least one of techniques using contextual keyword classifications, techniques using attention-based keyword classifier machine learning framework, techniques using a greedy matching indicator, and/or the like.