ML Metric Prediction for Service Routing Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional service environments face challenges in objectively determining the quality of service interactions due to incomplete and biased survey responses from customers, leading to inaccurate assessment of service quality and inefficient routing of service requests.
Innovation Solution
Implementing machine learning techniques, such as deep learning, to predict survey scores based on session records and cluster users, allowing for accurate routing of service requests to suitable service representatives without relying on complete surveys.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional survey-based service quality assessment is used, then service quality can be measured, but the measurements are incomplete and biased due to incomplete survey responses
Solution Approach 1:
The patent introduces machine learning models as intermediaries between session records and service quality assessment. These models process session records to generate predicted survey scores, serving as a mediator that translates objective session data into quality metrics without requiring actual survey responses from customers.
Solution Approach 2:
The patent replaces the mechanical survey-based assessment system with an automated machine learning-based prediction system. Instead of relying on customers to manually complete surveys, the system uses ML models to automatically predict survey scores from session records, eliminating the need for manual survey completion.
2Measurement precision
If machine learning models are trained on survey data, then predicted survey scores can be generated, but the system complexity increases
Solution Approach 1:
The patent applies preliminary action by training machine learning models in advance on historical survey data before deployment. The models are pre-trained to learn patterns between session records and survey scores, so that when deployed, they can directly predict scores without requiring complex real-time processing during actual service sessions.
Solution Approach 2:
The patent uses copying by training ML models on historical survey data to create a predictive representation. The models learn to replicate the survey scoring process by analyzing patterns in historical data, effectively creating a digital copy of the survey assessment process that can be applied to new sessions without actual surveys.
3Reliability
If users are clustered based on predicted metrics, then service request routing accuracy improves, but the routing system becomes more complex
Solution Approach 1:
The patent applies preliminary action by pre-clustering users based on their characteristics and predicted metrics before service requests arrive. The clustering is performed in advance using historical data and ML predictions, creating predefined user groups that can be quickly matched with appropriate service representatives when requests come in, avoiding complex real-time clustering decisions.
Solution Approach 2:
The patent uses segmentation by dividing users into distinct clusters based on their characteristics and predicted survey scores. This segmentation allows the routing system to match service requests with representatives specialized for specific user segments, improving routing accuracy while keeping the system manageable through organized groupings.
Data Source
AI summary
Techniques are described for generating metric(s) that predict survey score(s) for a service session. Model(s) may be trained, through supervised or unsupervised machine learning, using training data such as communications from previous service sessions between service representative(s) and individual(s), and survey scores provided by the serviced individual to rate the session on one or more criteria (e.g., survey questions). The model(s) may be trained to output, based on an input session record, metric(s) that each correspond to a survey score that would have been provided by the individual had they completed the survey. The model may be a concatenated model that combines a language model output from a language classifier recurrent neural network, and an acoustic model output from an acoustic feature layer convolutional neural network. Individuals can be clustered according to the metric(s) and/or other factors, and the cluster(s) can be employed for routing incoming service requests.