Conversational System Intent Mapping via Knowledge Graph Triplet Embedding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current conversational AI systems, such as IBM Watson Assistant, face inefficiencies and errors in managing intents due to the time-consuming and error-prone process of manually matching user utterances to intents, especially in medium and large-scale dialog systems.
Innovation Solution
The method involves embedding triplets of concepts and relationships into a knowledge graph, scanning call logs for examples, and mapping them to intents, while also creating and updating intents using a builder module that includes a concept manager and example finder to enhance scalability and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manually matching sentences in call logs to intents is used, then domain experts can define intents with examples, but the process is time-consuming and error-prone
Solution Approach 1:
The system automatically extracts triplets from call logs and maps them to intents without requiring manual intervention. The automated triplet extraction and intent mapping processes eliminate the time-consuming manual matching operation while maintaining accuracy through systematic processing of call log data
Solution Approach 2:
The patent replaces the manual mechanical process of domain experts reviewing and matching sentences with automated computational processes. The system uses automated triplet extraction algorithms and intent mapping mechanisms to substitute human labor, significantly reducing time requirements while improving consistency and reliability
2Adaptability or versatility
If a large collection of intents is used in medium and large-scale dialog systems, then system capability increases, but management complexity increases
Solution Approach 1:
The patent segments the intent management task by organizing intents into structured collections with hierarchical relationships. The system divides the management of large intent collections into manageable segments through structured data organization, enabling scalable management of medium and large-scale dialog systems without proportional increase in complexity
Solution Approach 2:
The patent introduces automated intermediary processes that mediate between the large collection of intents and the operational system. These intermediary automated processes handle the complexity of managing large intent collections by automatically extracting, mapping, and organizing triplets, thereby enabling system capability expansion without linear increase in management complexity
3Productivity
If automated triplet extraction and intent mapping is implemented, then efficiency improves, but system complexity increases
Solution Approach 1:
The patent implements a universal automated processing framework that handles multiple functions: triplet extraction from call logs, intent mapping, conflict detection, and example selection. This multi-functional system achieves high productivity by consolidating multiple operations into a single automated pipeline, managing complexity through functional integration rather than separate systems
Data Source
AI summary
A method of providing examples to a computerized conversation agent includes associating one or more triplets of two concepts and a relationship therebetween with an intent related to a query. The triplet is embedded in a knowledge graph and the concepts in the knowledge graph are mapped to the intent. A call log is scanned for examples of the intent based on the concepts in the knowledge graph and the examples are mapped to the intent.