Conference Call Poll Detection From Live Discussion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing conference call platforms require participants to manually introduce polling questions, which interrupts the natural flow and organization of discussions, increasing discussion length and resource utilization.
Innovation Solution
A machine learning model is trained to recognize polling questions from verbal phrases during a conference call, automatically designating them and displaying a message to the participant for confirmation, reducing interruptions and optimizing resource usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If participants pre-plan and interrupt the discussion to pose polling questions, then polling can be conducted, but the flow of discussion is disrupted and the call length increases
Solution Approach 1:
The system performs preliminary action by continuously analyzing speech patterns and identifying potential polling questions in real-time during the conference call, so that when a polling question is detected, the system can immediately present it to the participant for confirmation and execution without interrupting the discussion flow. This eliminates the need for participants to pre-plan polling questions and manually interrupt the discussion.
2Ease of operation
If participants pre-plan and interrupt the discussion to pose polling questions, then polling can be conducted, but system resource efficiency decreases
Solution Approach 1:
The system implements self-service by automatically detecting, analyzing, and presenting potential polling questions to participants without requiring manual intervention to initiate the polling process. The machine learning model continuously monitors the conversation and autonomously identifies polling opportunities, reducing the burden on participants and optimizing system resource utilization.
3Loss of time
If a machine learning model automatically identifies polling questions, then discussion flow is maintained, but model training and processing are required
Solution Approach 1:
The system performs preliminary action by training the machine learning model in advance with historical polling question data before deployment. This pre-training enables the model to accurately identify polling questions during conference calls with minimal processing time, reducing the trade-off between automation complexity and time savings.
Data Source
AI summary
Data indicating one or more verbal phrases provided by one or more participants during a conference call is fed as input to a machine learning model. One or more outputs of the machine learning model are obtained. A polling question for polling at least a portion of the participants is extracted from the one or more outputs of the machine learning model. The polling question is based on one or more verbal phrases provided by the one or more participants. The polling question is provided for polling the at least the portion of the participants during the conference call.