Insurance Knowledge Graph Reasoning for Accurate Multi-Round Q&A
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing intelligent question and answer systems are limited by their reliance on historical question and answer databases, requiring significant data and failing to provide accurate answers in scenarios without historical data, and are inadequate in handling multiple rounds of questions and answers.
Innovation Solution
A knowledge answering method and system that extracts insurance rules from documents, generates an insurance knowledge graph as a triplet network, and uses a reasoning engine to clarify user intent through multiple rounds of questioning, providing precise answers by searching the graph based on matched rules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing intelligent question and answer systems rely on historical question and answer databases, then they can provide answers based on past data, but they fail to provide accurate answers in scenarios without historical data and require significant data resources
Solution Approach 1:
The system segments knowledge into structured insurance rule triplets (subject, predicate, object) and organizes them into a knowledge graph. This segmentation allows the system to store and retrieve specific insurance rules without requiring extensive historical Q&A data, thereby improving answer accuracy while reducing data requirements.
Solution Approach 2:
The patent introduces a reasoning engine as an intermediary between the user question and the knowledge base. The reasoning engine extracts insurance rules from unstructured text, converts them into structured triplets, and queries the knowledge graph to find answers. This intermediary enables accurate answering without relying on historical Q&A databases.
2Adaptability or versatility
If existing systems use one round of question and answer, then the system structure is simple, but it is inadequate in handling multiple rounds of questions and answers
Solution Approach 1:
The system implements feedback mechanisms where the reasoning engine analyzes user questions, queries the knowledge graph, and returns answers that can prompt follow-up questions. This feedback loop enables multi-round interactions while maintaining manageable system complexity through the structured knowledge graph and rule-based reasoning.
Solution Approach 2:
The system dynamically adapts to multi-round conversations by maintaining context awareness through the reasoning engine, which can process sequential questions and update its understanding of user intent across multiple interactions, enabling versatile dialogue without excessive complexity.
3Measurement precision
If the system extracts and processes multiple insurance rules to generate a knowledge graph, then answer precision is improved, but the processing complexity and time increase
Solution Approach 1:
The system performs preliminary action by pre-extracting insurance rules from documents and organizing them into a structured knowledge graph before user queries arrive. This preprocessing creates ready-to-query triplets that enable fast, precise answer retrieval without time-consuming processing during actual Q&A sessions.
Solution Approach 2:
The patent transforms unstructured insurance text into structured triplet parameters (subject, predicate, object) with specific attributes. This parameter transformation enables efficient querying and precise matching during answer generation, improving answer precision while reducing processing time through standardized data formats.
Data Source
AI summary
A method for knowledge answering includes: receiving a user question entered by a client, and obtaining a target insurance rule matching the user question from a plurality of insurance rules; searching for an answer in an insurance knowledge graph based on the target insurance rule, in which the insurance knowledge graph is generated based on the plurality of insurance rules; and returning the answer to the client.