Question-Answering System Using Knowledge Graph Context
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing question-answering systems face reliability issues due to insufficient keywords in questions leading to incorrect answer candidates and low reliability of inferred answers.
Innovation Solution
A customized question-answering system based on a knowledge graph that includes a user interface, natural language understanding unit, integrated knowledge graph management, and question target ranking unit to analyze and classify questions, generate context knowledge graphs, and provide accurate answers or reverse questions by selecting relevant entity knowledge graphs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing question-answering systems use simple keyword-based document searches, then the system complexity is low, but the answer reliability is insufficient due to insufficient keywords leading to incorrect answer candidates
Solution Approach 1:
The patent introduces a knowledge graph as an intermediary component between the question input and answer generation. The knowledge graph management unit extracts entities and relationships from the question, builds a context knowledge graph, and queries it to obtain structured knowledge. This intermediary structure enables more reliable answer generation by leveraging predefined relationships and entities, rather than relying solely on keyword matching in unstructured documents.
Solution Approach 2:
The system segments the question-answering process into distinct functional modules: natural language understanding unit for question analysis, knowledge graph management unit for entity extraction and context graph generation, and question target ranking unit for answer selection. This segmentation allows each component to specialize in specific tasks, improving overall answer reliability while maintaining manageable system complexity through modular design.
2Adaptability or versatility
If the system generates numerous correct answer candidates through document searches, then the coverage of possible answers is improved, but the difficulty of detecting and measuring the correct answer increases due to ranking and inference requirements
Solution Approach 1:
The system performs preliminary action by pre-building a knowledge graph that contains structured entities and relationships before the question-answering process. The knowledge graph management unit extracts entities, attributes, and relationships from available documents in advance, organizing them into a structured format with defined schemas. This preliminary structuring makes it easier to detect and measure correct answers during query processing, as the system can directly query predefined relationships rather than searching unstructured text.
3Productivity
If the system infers answers with the highest ranking, then the answer generation speed is improved, but the reliability of inferred answers decreases due to potential incorrect inferences
Solution Approach 1:
The system implements feedback mechanisms through the question target ranking unit, which evaluates answer candidates by comparing them against the context knowledge graph. The ranking process considers multiple factors including entity matches, relationship consistency, and attribute alignment with the question. This feedback loop ensures that high-ranking answers are not only generated quickly but are also validated against structured knowledge, improving reliability while maintaining productivity.
Data Source
AI summary
This application relates to a user-customized question-answering system based on a knowledge graph that can answer questions accurately. In one aspect, the system includes a user interface configured to receive a question or a reply in the form of a natural language from a user and transmit an answer or reverse question to the user, through a network, and a natural language understanding unit configured to perform natural language analysis and question classification. The system may also include an integrated knowledge graph configured to store or manage a knowledge graph including multi-order relationship triples and a knowledge graph management unit configured to generate a context knowledge graph having entity knowledge graphs and select at least one entity knowledge graph from among the entity knowledge graphs. The system may further include a question target ranking unit configured to generate the answer or the reverse question for the question.