Action Item Log Generation Using ML Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating action item logs from document data objects are inefficient, leading to non-harmonious content segments, increased user burden, and high operational loads on document collaboration server systems due to inefficient text processing and inaccurate action item extraction.
Innovation Solution
The use of action item classification and extraction machine learning models, including a part-of-speech tagger and sequence processing model, to parse content data into segmentation units, predict action item presence, and generate concise action item logs, reducing the computational time and improving training efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional text processing methods are used to generate action item logs, then the system can process documents, but the processing time is excessive and operational load on server systems is high
Solution Approach 1:
The patent segments documents into content segmentation units (CSUs) at the sentence or paragraph level, enabling parallel processing of multiple segments simultaneously. This segmentation allows the system to process documents more efficiently by dividing the large-scale text processing task into smaller, manageable units that can be handled independently and in parallel, significantly reducing overall processing time and server operational load.
2Measurement precision
If comprehensive text processing is performed on all content segments, then action item extraction may be thorough, but computational resources are wasted on segments that do not contain action items
Solution Approach 1:
The patent implements a two-stage processing approach where a classification model first performs preliminary action to identify and filter CSUs that are likely to contain action items. Only the filtered subset of CSUs with high probability of containing action items is then passed to the extraction model for detailed processing. This preliminary filtering step significantly reduces computational resource consumption by avoiding exhaustive processing of all document segments while maintaining high extraction accuracy for relevant content.
3Measurement precision
If manual review of document content is required to identify action items, then extraction accuracy may be high, but user burden increases and processing efficiency decreases
Solution Approach 1:
The patent implements self-service through automated machine learning models that independently perform action item identification and extraction without requiring manual user review. The classification and extraction models work autonomously to analyze document content, identify action items, and generate structured outputs. This automation eliminates the need for users to manually review and identify action items, significantly reducing user burden while maintaining high accuracy through the sophisticated ML models.
4Quantity of substance
If detailed processing is applied to all content segmentation units, then action item extraction may be comprehensive, but the system complexity and processing overhead increase
Solution Approach 1:
The patent reduces system complexity and processing overhead by implementing preliminary filtering through a classification model that identifies only those CSUs likely to contain action items. This filtering step creates a reduced subset of CSUs that requires detailed processing, thereby simplifying the overall processing pipeline and reducing the number of units that need comprehensive analysis. The system maintains comprehensive action item extraction by ensuring the classification model accurately identifies relevant segments before extraction.
Solution Approach 2:
The patent segments the processing task into two distinct phases: classification (filtering) and extraction (detailed processing). This segmentation allows the system to apply different levels of processing intensity to different segments, reducing overall system complexity by avoiding uniform detailed processing of all CSUs. The segmented approach enables efficient resource allocation and simplifies the processing architecture by separating the filtering function from the extraction function.
Data Source
AI summary
Various embodiments of the present invention provide methods, apparatuses, systems, computing devices, and/or the like that are configured to accurately and concisely generate one or more action item logs of one or more document data objects. For example, certain embodiments of the present invention provide methods, apparatuses, systems, computing devices, and/or the like that are configured to generate an action item log of a document data object comprising one or more semantically complete or incomplete units of text data, by generating content segmentation units, determining action item presence predictions, generating action item sets from each content segmentation unit within a candidate action item subset, aggregating the action item sets to create an action item log, and storing the action item log.