Speech and Sentiment Feature Engineering for Predictive Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional skill-based routing in contact centers is static and requires manual effort to maintain, failing to leverage new data types and adapt dynamically to real-time changes, leading to suboptimal customer-agent matching.
Innovation Solution
A method for predictive routing that processes interaction data, including audio and transcript analysis, to generate features for a machine learning model that optimally matches customers with agents based on speech and sentiment metrics, reducing manual effort and enhancing adaptability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional skill-based routing with explicit skill models is used, then routing decisions can be made based on agent capabilities, but the models are static and require manual effort to construct and maintain
Solution Approach 1:
The system automatically generates and updates skill models by processing interaction data, audio data, and transcript data through speech analytics and sentiment analysis. The model dynamically adapts to real-time changes without manual intervention, allowing the system to self-service the complex task of model construction and maintenance while improving routing accuracy
Solution Approach 2:
Manual model construction and maintenance is replaced by an automated machine learning system that processes data and generates routing models. The mechanical process of manually updating skill models is substituted with an automated computational system that continuously learns from new data, reducing complexity while maintaining reliability
2Adaptability or versatility
If traditional skill-based routing is used, then routing decisions can be made, but the system fails to leverage new data types and adapt dynamically to real-time changes
Solution Approach 1:
The routing system transitions from static skill models to dynamic models that continuously adapt to real-time changes. The system processes new interaction data, audio recordings, and transcripts to update routing decisions dynamically, enabling the model to respond to changing conditions while fully utilizing available data types
Solution Approach 2:
The system integrates multiple data types (interaction data, audio data, transcript data) and processing methods (speech analytics, sentiment analysis) into a unified routing framework. This multi-functional approach allows the system to leverage diverse data sources simultaneously, improving adaptability without losing information from any single data type
Data Source
AI summary
A method for processing data for training a predictive routing model. The method includes receiving interaction data from previous interactions that includes audio data capturing a conversation and transcript data of the conversation. The method continues by performing speech analytics by processing the audio data to determine scores for speech metrics that include a measure of how much the agent or customer speaks during the conversation. The method continues by performing sentiment analysis to determine scores associated with sentiment metrics, the sentiment metrics including a measure of a sentiment based on classifying utterances appearing in the transcript data as being positive or negative. The method continues by performing feature engineering to generate feature data and generating a training dataset therefrom. The method continues by applying a machine learning algorithm to the training dataset to train a predictive routing model.