Semantic Frame Builder for IVR Intent Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Interactive voice response (IVR) systems face challenges in accurately identifying the intent of human inquiries and responding appropriately, particularly in classifying components of inquiries based on industry-specific semantic classifiers and understanding semantic relationships within the inquiries.
Innovation Solution
The development of a semantic frame building system that uses a machine learning algorithm to index and classify tokens in an utterance, assign semantic roles, and build semantic frames to identify intent and summarize conversations, enabling IVR systems to provide accurate responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If IVR systems use traditional intent identification methods, then the system complexity is low, but the accuracy of identifying human inquiry intent and classifying components based on industry-specific semantic classifiers is insufficient
Solution Approach 1:
The patent segments the utterance into multiple tokens and classifies each token individually using semantic role classifiers. This segmentation allows the system to handle complex inquiries by processing smaller units, improving overall accuracy while managing complexity through systematic breakdown of the classification task.
Solution Approach 2:
The patent introduces a semantic frame dimension to traditional intent classification. By creating semantic frames that capture relationships between tokens and their roles (e.g., subject, object, action), the system adds a new dimension of analysis that improves intent identification accuracy without simply increasing the number of classification categories.
2Adaptability or versatility
If the classification module uses a limited set of semantic classifiers corresponding to industry categories, then the ease of operation and industry-specific relevance are improved, but the ability to handle diverse and complex inquiries is reduced
Solution Approach 1:
The patent creates a universal semantic frame structure that can be applied across different industry categories. The same semantic frame template (with variables for different entities and relationships) serves multiple industries, allowing the system to handle diverse inquiries while maintaining ease of operation through a standardized classification approach.
Solution Approach 2:
The system uses parameter changes by adjusting the semantic frame variables to match different industry contexts. The same semantic frame structure can accommodate different industries by changing the specific entities and relationships parameters, providing versatility without requiring separate classification systems for each industry.
3Measurement precision
If the system indexes all tokens and assigns semantic roles to every token, then the measurement precision of semantic relationships is improved, but the processing time and computational resources are increased
Solution Approach 1:
The patent applies partial action by not necessarily processing every token in every possible detail, but rather focusing on indexing and assigning semantic roles to tokens that are most relevant for intent identification. This selective processing maintains sufficient measurement precision for semantic relationships while reducing unnecessary computational time.
Data Source
AI summary
Systems are provided for building semantic frames. Systems may include building a semantic frame using a machine learning algorithm. The algorithm may identify: an index number of a token, a semantic role classifier assigned to the token, a corresponding correlation value and an index number of one or more related tokens. The algorithm may also create a semantic frame using the identified information. Systems may include building semantic frames for multiple tokens within an utterance. Systems may include building semantic frames for a plurality of tokens within a plurality of utterances. The plurality of utterances may be components of a conversation. Systems may also include summarizing the conversation using the semantic frames.