Hybrid Deep Learning Routing Model for Customer Service
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Software platforms face challenges in efficiently routing user requests to the most appropriate customer service agents due to non-intuitive product and service offerings, leading to increased handling times and operational costs.
Innovation Solution
A hybrid model combining deep learning and business logic is used to generate routing decisions based on user input features and context, optimizing the routing of customer service requests to the most suitable support queues and agents.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If user requests are routed to customer service agents using traditional methods, then routing decisions can be made with simple rules, but routing accuracy deteriorates leading to increased handling times
Solution Approach 1:
The patent transforms the routing decision process by changing from simple rule-based parameters to multi-channel deep learning model parameters that analyze multiple features including user input, context, and request characteristics. This enables more accurate routing predictions while maintaining operational efficiency.
Solution Approach 2:
The patent employs a hybrid model that combines multiple deep learning models (BERT, RoBERTa, DistilBERT) with business logic rules, creating a composite routing system that leverages the strengths of both approaches to achieve high routing accuracy without sacrificing handling efficiency.
2Measurement precision
If deep learning models are used to improve routing accuracy, then routing precision improves, but compute resource requirements increase
Solution Approach 1:
The patent implements a multi-channel model selection approach where simpler models (DistilBERT) are used for straightforward routing decisions, while more complex models (BERT, RoBERTa) are selectively applied only when needed for difficult-to-route requests. This partial application of computational resources maintains high routing accuracy while reducing overall compute consumption.
Solution Approach 2:
The routing system is segmented into multiple independent deep learning models, each optimized for specific routing scenarios. This segmentation allows the system to deploy only the necessary computational resources for each individual routing decision rather than always using the most computationally intensive model.
3Adaptability or versatility
If multiple deep learning models are deployed to handle diverse routing scenarios, then routing versatility improves, but device complexity increases
Solution Approach 1:
The patent designs the multi-channel deep learning architecture to serve multiple routing functions simultaneously. The same model framework handles different request types, user contexts, and routing scenarios through a unified interface, reducing the operational complexity despite the diversity of models deployed.
Solution Approach 2:
The patent introduces a model selection layer that acts as an intermediary between the multiple deep learning models and the routing decision process. This intermediary manages model deployment, selects appropriate models based on request characteristics, and coordinates their outputs, thereby simplifying the overall system complexity while maintaining versatility.
Data Source
AI summary
Systems and methods are used to generate contact type predictions that route user customer service requests within a support platform. The contact type predictions are generated using a hybrid model that includes a deep learning component and a business logic component. The deep learning component may generate a multi-channel output based on text features and context features. The multi-channel output is modified based on one or more business rules to generate the contact type predictions.