Dialog Processing System for Mandatory Utterance Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for extracting mandatory utterances in sales transactions and compliance checks from business conversations are inefficient due to the need for manual labeling and industry-specific data preparation, leading to variations and high resource requirements.
Innovation Solution
A dialog processing system that extracts mandatory utterances by using field-independent utterance patterns and feature quantities from general conversations, converting them into text format, and generating relaxed patterns to accurately identify mandatory data without prior knowledge of the industry or application.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual labeling is used to prepare industry-specific data for speech recognition, then the speech recognition system can be trained for specific applications, but the work requires extensive manual effort and produces variations in labeling ranges
Solution Approach 1:
The system performs preliminary extraction of mandatory utterances using automatically generated patterns before final speech recognition processing. By pre-identifying and extracting mandatory utterance portions using the pattern matching unit and feature quantity extraction, the system reduces the need for extensive manual labeling while maintaining recognition accuracy for compliance-critical content
Solution Approach 2:
The system uses its own output (extracted mandatory utterances and their feature quantities) to automatically generate and refine patterns for future extraction. The pattern generation unit creates patterns based on extracted feature quantities, enabling the system to self-improve without external manual intervention for each new industry or application
2Adaptability or versatility
If manual labeling is performed for each industry and application, then the speech recognition system can be customized, but the process requires repeated manual work and high resources
Solution Approach 1:
The system extracts feature quantities from mandatory utterances across different industries and applications, identifying universal patterns that transcend specific domains. The extracted feature quantities capture essential characteristics of mandatory utterances (such as confirmation requests, risk disclosures, and compliance statements) that are common across insurance, finance, healthcare, and other regulated industries, enabling a single system to serve multiple purposes without industry-specific relabeling
Solution Approach 2:
The system changes the parameter representation from industry-specific labeled data to universal feature quantities derived from utterance patterns. By transforming mandatory utterances into standardized feature quantity representations (such as temporal patterns, linguistic structures, and semantic characteristics), the system enables cross-industry applicability while maintaining the ability to adapt to specific regulatory requirements through pattern adjustment rather than complete relabeling
3Extent of automation
If speech recognition technology is used to check mandatory utterances, then automation is achieved, but previous knowledge regarding data and fields of application must be prepared manually
Solution Approach 1:
The system extracts and isolates the essential characteristics of mandatory utterances through feature quantity extraction from pattern-matched content. By separating the core identifying features of mandatory utterances (such as specific temporal patterns, linguistic markers, and structural characteristics) from the rest of the conversation data, the system creates a simplified representation that can be automatically recognized without requiring complex industry-specific knowledge bases
Solution Approach 2:
The system performs preliminary extraction of mandatory utterances using automatically generated patterns before final speech recognition processing. By pre-identifying and extracting mandatory utterance portions using the pattern matching unit and feature quantity extraction, the system reduces the need for extensive manual labeling while maintaining recognition accuracy for compliance-critical content
Data Source
AI summary
A dialog processing system which includes a target expression data extraction unit for extracting a plurality of target expression data each including a pattern matching portion which matches an utterance pattern, which are inputted by an utterance pattern input unit and is an utterance structure derived from contents of field-independent general conversations, among a plurality of utterance data which are inputted by an utterance data input unit and obtained by converting contents of a plurality of conversations in one field; a feature extraction unit for retrieving the pattern matching portions, respectively, from the plurality of target expression data extracted, and then for extracting feature quantity common to the plurality of pattern matching portions; and a mandatory data extraction unit for extracting mandatory data in the one field included in the plurality of utterance data by use of the feature quantities extracted.