Conversation Path Classification for Natural-Language IVR
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing IVR systems lack a natural, conversational interface, leading to user frustration and inefficiency, while computer-based chat systems and virtual assistants use unstructured natural language input incompatible with IVR formats.
Innovation Solution
A computer-based system classifies conversation segments by identifying dominant paths in transcribed conversations, creating a conversation model that enhances IVR systems with a conversational interface, optimizing resource use and reducing delays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If IVR systems use structured menu interfaces, then system control and predictability are improved, but user convenience and natural interaction are worsened
Solution Approach 1:
The patent introduces a conversation model as an intermediary layer between the user's natural language input and the IVR system's structured menu hierarchy. This model translates unstructured conversational segments into structured path traversals, enabling the system to maintain control while accepting natural user interaction styles
Solution Approach 2:
The system changes the input format parameter from structured menu selections to unstructured natural language, then transforms it into a standardized internal representation (conversation segments with paths) that the IVR system can process. This parameter transformation allows both natural interaction and system control to coexist
2Device complexity
If IVR systems use traditional menu hierarchies, then system complexity is reduced, but interaction time and user frustration increase
Solution Approach 1:
The patent segments the conversation into discrete segments with identified paths between conversation turns. This segmentation allows the system to process complex conversations by breaking them into manageable units, reducing the cognitive load on the system while maintaining fast interaction through automated path recognition
Solution Approach 2:
The system performs preliminary classification of conversation segments into dominant paths before full processing. By pre-identifying the most likely conversation flow paths, the system can optimize processing efficiency and reduce interaction time without increasing complexity
3Ease of operation
If chat systems use unstructured natural language, then ease of operation is improved, but compatibility with traditional IVR systems deteriorates
Solution Approach 1:
The conversation model serves as a bridge between unstructured natural language chat interfaces and structured IVR systems. It receives natural language input, classifies it into conversation segments, and outputs standardized path traversals that are compatible with traditional IVR processing
Solution Approach 2:
The system creates a universal conversation model that can handle both unstructured natural language input from chat interfaces and structured path output for IVR compatibility. This multi-functional model enables seamless integration between different interaction paradigms
Data Source
AI summary
Computer-based natural language understanding of input and output for a computer interlocutor is improved using a method of classifying conversation segments from transcribed conversations. The improvement includes identifying, by a computer processor, a dominant path of conversational behavior within conversation text data, wherein the dominant path comprises a plurality of path segment traversals between conversation turns in the conversation text data, and wherein the plurality of path segment traversals are accumulated into a weight for the one or more dominant paths; and modifying, by a computer processor, a digital conversation model in a computer-readable non-transitory memory device to contain the identified dominant path.