Hybrid Semantic Model for Accurate Call Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech recognition systems in automated call routing often misinterpret customer intent, leading to inefficient call routing and increased customer dissatisfaction due to misrouting, resulting in significant operational costs and abandoned calls.
Innovation Solution
A hybrid semantic model is integrated with action-object technology, where speech input is converted into text and matched against action-object pairs with confidence levels, allowing for accurate routing by selecting dominant actions and complementary objects, potentially approaching 100% correct destination assignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional speech recognition systems are used for automated call routing, then the system is simple to operate, but the call routing accuracy deteriorates leading to misrouting
Solution Approach 1:
The speech recognition system is divided into multiple independent modules: acoustic model for phoneme recognition, semantic model for intent understanding, and action-object model for task identification. Each module processes specific aspects of speech independently, then results are integrated to achieve high routing accuracy without overwhelming system complexity
Solution Approach 2:
A semantic model acts as an intermediary layer between the acoustic model and the call routing system. It translates raw speech patterns into meaningful intent representations, enabling accurate routing decisions while maintaining clear separation between recognition and routing functions
2Ease of operation
If speech recognition systems attempt to interpret natural language intent, then customer satisfaction improves, but misinterpretation errors increase leading to misrouting
Solution Approach 1:
The system transforms the interpretation task from understanding complete natural language sentences to identifying specific action-object pairs with associated confidence levels. This parameter transformation allows the system to handle natural language flexibility while maintaining reliable, objective decision criteria for routing
Solution Approach 2:
The system replaces subjective human interpretation of natural language with an automated confidence-based selection mechanism. The action-object model with confidence scoring objectively determines the most likely customer intent, eliminating human bias and inconsistency while maintaining ease of use
3Productivity
If the system routes calls automatically without human intervention, then productivity increases, but the confidence level in routing decisions deteriorates
Solution Approach 1:
The system performs partial human intervention by automatically routing high-confidence calls while flagging low-confidence calls for human review. This partial automation maintains high productivity for routine calls while preserving human judgment for ambiguous cases, achieving both efficiency and reliability
Data Source
AI summary
A method for processing a call is disclosed. The method receives a speech input in connection with a call and transforms at least a segment of the speech input into a first textual format. The method also generates a first list of entries based, at least partially, on a consideration of the first textual format, the first list comprising at least one action with a corresponding confidence level and at least one object with another corresponding confidence level, selects an entry of the first list having a higher corresponding confidence level, outputs a second textual format. The method further generates a second list based, at least partially on consideration of the selected entry and the second textual format and suggesting a routing option for the call based on the selected entry and a pairing entry in the second list having a high corresponding confidence level.