Semantic Classifier Knowledge Reuse for Call Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing call routing systems require a large, expensive, and domain-specific training corpus for each new application, making it difficult to reuse knowledge across different applications and domains, and face challenges in scaling due to exponential growth of joint outputs from multiple classifiers.
Innovation Solution
The method involves using a pre-existing semantic classifier and its classification tags to tag and train a new semantic classification application, allowing for a soft-mapping of knowledge across different applications and domains, and abstracting application-specific features into generic stem rules to create a new semantic classifier, which can handle different classification tags and domains efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a new training corpus is developed for each new call routing application, then the classification accuracy for that specific application is improved, but the development time and cost increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-tagging utterances in the new training corpus using the pre-existing semantic classifier and its classification tags before actual training begins. This preliminary tagging step reuses existing classification knowledge, eliminating the need to create classification tags from scratch for each new application, thereby reducing development time while maintaining classification accuracy.
Solution Approach 2:
The patent applies copying by reusing the pre-existing semantic classifier and its classification tags from an earlier application and applying them to the new application. Instead of creating entirely new classification systems, the patent copies and adapts existing classification knowledge to new domains, significantly reducing development effort while preserving accurate classification performance.
2Measurement precision
If multiple classifiers are combined to improve classification performance, then the accuracy is improved, but the system complexity and scaling become difficult due to exponential growth of joint outputs
Solution Approach 1:
The patent applies universality by creating a generic semantic classifier that can function across multiple different applications and domains. Instead of combining multiple specialized classifiers for different functions, the patent develops a single universal classifier that can handle various classification tasks by reusing and adapting its classification tags, thereby avoiding the exponential complexity growth associated with combining multiple specialized classifiers.
Solution Approach 2:
The patent applies parameter changes by allowing the semantic classifier to adapt its classification tags and parameters to different applications and domains. Rather than fixing the classifier structure and combining multiple rigid classifiers, the patent enables dynamic parameter adjustment and tag reuse across different contexts, simplifying the system while maintaining accuracy across diverse applications.
Data Source
AI summary
A method is described for semantic classification in human-machine dialog applications, for example, call routing. Utterances in a new training corpus of a new semantic classification application are tagged using a pre-existing semantic classifier and associated pre-existing classification tags trained for an earlier semantic classification application.