Machine Learning Polling-Question Detection for Conference Calls
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing conference call platforms face inefficiencies due to interruptions and increased length caused by participants manually posing polling questions, which disrupt the natural flow and require significant system resources, leading to reduced efficiency and increased latency.
Innovation Solution
A machine learning model is trained to recognize polling questions from verbal phrases during a conference call, automatically designating them and populating answers based on initial responses, reducing the need for manual intervention and minimizing discussion interruptions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If participants manually pose polling questions during a conference call, then the polling function can be performed, but the discussion flow is interrupted and the call length increases
Solution Approach 1:
The system automatically detects and processes polling questions from participant speech without requiring manual intervention. The machine learning model identifies polling questions, extracts answer options, and manages the polling process autonomously, allowing participants to simply speak their questions naturally during the conference call.
2Productivity
If participants manually pose polling questions, then polling can be conducted, but system resource usage increases and latency increases
Solution Approach 1:
The patent replaces manual mechanical operations (participants manually typing or selecting polling questions from interfaces) with an automated machine learning-based speech recognition system. This substitution reduces the computational overhead of manual interface interactions and eliminates the latency associated with manual input processing.
3Ease of operation
If a machine learning model automatically recognizes polling questions, then discussion flow is maintained and efficiency improves, but device complexity increases
Solution Approach 1:
The machine learning model is designed to perform multiple functions: transcribing speech, identifying polling questions, extracting answer options, and managing the polling process. This multi-functionality consolidates what would otherwise require separate systems into a single unified model, reducing overall system complexity despite the advanced capabilities required.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Systems and methods for recognizing polling questions from a conference call discussion are provided. One or more text strings including a textual form of one or more verbal phrases provided by one or more participants of a conference call are identified. The one or more text strings are provided as input to a trained machine learning model. One or more outputs from the trained machine learning model are obtained. A level of confidence that a verbal phrase of the one or more verbal phrase provided by the one or more participants includes a question associated with polling during the conference call is extracted from the one or more outputs. In response to a determination that the level of confidence satisfies a confidence criterion, the verbal phrase is designated as a polling question presented during the conference call.