Temporal Emotion Analysis in Contact Center Audio
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Previous methods for analyzing business conversations in contact centers primarily detect global emotions without analyzing the temporal evolution of emotions or the factors influencing these changes, failing to provide insights on converting negative sentiment to positive sentiment.
Innovation Solution
A computer-implemented method that identifies changes in emotions during audio interactions between contact center agents and consumers, analyzing these changes in conjunction with various aspects of the interaction to determine the relationship between emotional shifts and specific factors, such as hold time, agent empathy, and customer history, to improve sentiment conversion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If global emotion detection is used over entire conversations, then emotion classification is achieved, but temporal evolution and influencing factors are lost
Solution Approach 1:
The patent segments the conversation into multiple time windows or segments, analyzing emotion in each segment separately rather than treating the entire conversation as a single unit. This allows tracking of temporal evolution while maintaining classification accuracy in each segment.
Solution Approach 2:
The patent adds a temporal dimension to emotion analysis by examining emotions across multiple time points and correlating them with events occurring at specific times. This transforms the analysis from a single global classification to a multi-dimensional temporal analysis that preserves evolutionary information.
2Measurement precision
If emotion is detected in isolation, then emotion classification is achieved, but insights into what led to emotion changes are lost
Solution Approach 1:
The patent incorporates feedback by analyzing the relationship between emotions and conversational events, using this information to understand what causes emotion changes. This feedback loop enables identification of causal relationships between specific events and emotional responses.
Solution Approach 2:
The patent merges emotion detection with event detection and analysis by simultaneously identifying both emotions and relevant conversational events, then analyzing their relationships. This combination preserves information about what led to emotion changes while maintaining classification accuracy.
3Loss of information
If detailed temporal emotion analysis is performed, then insights into emotion changes are achieved, but computational complexity increases
Solution Approach 1:
By segmenting the conversation into manageable time windows, the patent reduces the computational complexity of analyzing temporal evolution. Each segment can be processed independently, making the overall complex analysis tractable through divide-and-conquer.
Solution Approach 2:
The patent focuses analysis on relevant segments and events rather than processing every moment of the conversation in equal detail. By identifying and focusing on key moments where emotion changes occur, the system achieves detailed temporal analysis without excessive computational complexity.
Data Source
AI summary
Analyzing an audio interaction is provided. At least one change in an emotion of a speaker in an audio interaction and at least one aspect of the audio interaction are identified. The at least one change in an emotion is analyzed in conjunction with the at least one aspect to determine a relationship between the at least one change in an emotion and the at least one aspect, and a result of the analysis is provided.