Knowledge Graph Summaries for Efficient Domain Question Answering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing natural language generation systems, particularly in question-answering (QA) systems, face challenges due to high computational resource requirements, limited domain-specific information, and inefficiencies in retrieving relevant information, leading to suboptimal answer quality and resource consumption.
Innovation Solution
A knowledge graph incorporating textual summaries is used to provide improved input context to language models, enabling more efficient retrieval and generation of answers by organizing information in a tree format with document summaries at the root and textual passages at the leaves, reducing the need for storing entire documents and minimizing computational and memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If entire documents are stored and processed in natural language generation systems, then complete information is available for answer generation, but computational resource requirements and memory consumption increase significantly
Solution Approach 1:
The patent extracts only the necessary textual passages and summaries from complete documents that are relevant to the given question. Instead of processing entire documents, the system retrieves and processes only the specific portions containing relevant information, thereby reducing computational resource consumption while maintaining information completeness for answer generation.
Solution Approach 2:
The patent segments documents into smaller textual passages and hierarchical summaries (document-level, section-level, and paragraph-level summaries). This segmentation allows the system to process and store only the necessary granular units rather than complete documents, reducing memory requirements and computational overhead while preserving access to complete information when needed.
2Reliability
If domain-specific information and proprietary data are incorporated into QA systems, then answer quality improves, but system complexity and data management requirements increase
Solution Approach 1:
The patent creates a universal knowledge graph structure that can accommodate multiple types of information sources including domain-specific information and proprietary data. The hierarchical summary structure and retrieval mechanism work uniformly across different data types, allowing the system to incorporate diverse information sources without proportionally increasing system complexity.
3Productivity
If traditional QA systems retrieve relevant information, then answer generation can proceed, but retrieval efficiency is insufficient leading to suboptimal answer quality
Solution Approach 1:
The patent performs preliminary action by pre-computing and storing hierarchical summaries (document-level, section-level, and paragraph-level) for all documents in the knowledge graph before query processing. When a question is asked, the system can quickly retrieve pre-prepared summaries and relevant passages without performing complex real-time analysis, thereby improving retrieval efficiency while maintaining high answer quality through access to comprehensive pre-processed information.
Data Source
AI summary
Techniques are provided for producing an answer to a question regarding a domain. A natural-language textual sequence representing the question is received. From a knowledge graph associated with the domain, first and second textual passages are received using rankings corresponding to the natural-language textual sequence, a first textual summary is received summarizing textual information in a first vicinity of the first textual passage, and a second textual summary summarizing textual information in a vicinity of the second textual passage is received. An answer to the question is obtained using a language model by encoding a first intermediate output based on the natural-language textual sequence, the first textual passage, and the first textual summary, encoding a second intermediate output based on the natural language textual sequence, the second textual passage, and the second textual summary, and decoding a concatenation of the first and second intermediate outputs. An output is provided.