Agenda Coverage Prediction Using Topic Detection in Meetings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital communication platforms lack the ability to dynamically predict agenda item coverage during a meeting, requiring manual review and potential human error in determining if agenda items have been covered, which is time-consuming and prone to inaccuracies.
Innovation Solution
A system that classifies agenda items as long or short, uses topic detection models for long items and matching methods for short items to predict coverage based on utterances, and transmits the status of agenda items in real time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual review is used to determine agenda item coverage, then accuracy can be maintained through human judgment, but time consumption increases and human error may occur
Solution Approach 1:
The patent replaces the manual mechanical review process with an automated computer-based system that uses natural language processing and machine learning algorithms to analyze meeting transcripts and determine agenda item coverage. This substitution eliminates human time consumption and potential errors while maintaining or improving determination accuracy through consistent automated evaluation criteria.
Solution Approach 2:
The system enables the meeting documentation process to self-evaluate agenda item coverage automatically. The computer system independently analyzes the meeting transcript, compares it against predefined agenda items, and generates coverage determinations without requiring external manual review, thereby freeing up time while maintaining accuracy through automated assessment.
2Productivity
If automated prediction is implemented, then time efficiency improves and manual review is reduced, but system complexity increases
Solution Approach 1:
The patent implements a multi-functional computer system that not only predicts agenda item coverage but also transmits notifications to participant devices and manages the overall meeting documentation workflow. By consolidating these functions into a single automated system, the patent achieves high productivity while managing complexity through integrated design rather than multiple separate systems.
Solution Approach 2:
The system introduces a computer-based intermediary that acts as a mediator between the meeting transcript and the agenda item coverage determination. This intermediary layer processes the transcript through natural language processing and machine learning models, providing an automated prediction mechanism that improves productivity while containing complexity within a dedicated processing layer rather than requiring complex human workflows.
3Speed
If real-time prediction is provided, then immediate feedback is available to participants, but processing requirements and system complexity increase
Solution Approach 1:
The patent performs preliminary classification of agenda items into long and short categories before the actual prediction process. This preliminary action allows the system to apply different processing strategies - using topic detection models for long items and matching methods for short items - thereby achieving real-time prediction speed while managing complexity through pre-established processing pathways rather than uniform complex analysis for all items.
Solution Approach 2:
The system segments the agenda item prediction process into distinct pathways based on item characteristics. Long agenda items are processed through topic detection models that analyze meeting content in detail, while short agenda items use faster matching methods. This segmentation enables real-time notification speed by optimizing processing complexity according to the specific requirements of each agenda item type rather than applying uniform complex processing to all items.
Data Source
AI summary
A system classifies each agenda item as a long item or a short item; for agenda items classified as long items, extracts one or more topics from the utterances, and uses a topic detection model to predict whether a topic related to each agenda item classified as a long item has been covered; for agenda items classified as short items, applies one or more matching methods to predict whether one or more of the utterances within the sentence threshold cover each agenda item classified as a short item; and transmits, to one or more client devices, a status of the agenda items for the communication session.