Knowledge Base Taxonomy for Conversational Agent Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital conversational agents face challenges in acquiring and representing domain knowledge effectively, particularly in understanding customer expressions and providing accurate information, as they lack the ability to be audited for knowledge accuracy, leading to potential errors in customer service interactions.
Innovation Solution
A system and method for automatically generating a knowledge base taxonomy from structured or unstructured computer text to enhance the domain knowledge of conversational agents, utilizing a taxonomy generation module that extracts relevant terms from historical user interactions, organizes them into a semantic structure, and provides this structure to the agents for improved response generation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If digital conversational agents use extracted knowledge from textual sources, then they can acquire domain knowledge, but the knowledge cannot be audited by businesses leading to potential errors
Solution Approach 1:
The patent introduces an intermediary auditing layer between the conversational agent and the knowledge base. This layer includes approval workflows where business users can review, approve, or reject knowledge items before they are used by the agent. The intermediary mechanism allows automated knowledge extraction while maintaining human oversight to ensure accuracy and compliance.
Solution Approach 2:
The knowledge representation is segmented into structured components including knowledge items, concepts, and hierarchical relationships. Each knowledge item is divided into discrete elements that can be individually audited, approved, and tracked. This segmentation enables fine-grained control and auditing of knowledge accuracy without requiring complete manual review of the entire knowledge base.
2Reliability
If manual knowledge curation is used to ensure accuracy, then knowledge reliability improves, but the process becomes time-consuming and cannot scale
Solution Approach 1:
The system performs preliminary automated extraction and structuring of knowledge from textual sources before the auditing phase. Knowledge items are pre-processed, validated against existing knowledge bases, and organized into structured formats in advance. This preliminary action reduces the manual workload during auditing while maintaining high accuracy standards.
Solution Approach 2:
The system implements feedback loops where auditing outcomes are used to improve future knowledge extraction. Approved knowledge items are added to the knowledge base and used to refine extraction algorithms. Rejected items provide feedback for adjusting extraction parameters. This continuous feedback mechanism increases both accuracy and efficiency over time.
Data Source
AI summary
Methods and apparatuses are described for building a knowledge base taxonomy from structured or unstructured computer text for use in automated user interactions. A server computing device receives one or more of structured text or unstructured text corresponding to historical user interaction data from a database. The server computing device extracts one or more terms from the received text that are most relevant to a subject matter domain. The server computing device organizes the extracted one or more terms into a taxonomy data structure.