Knowledge Graph Summaries for Low-Resource Answer Generation
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 generated to provide improved input context to language models, utilizing less resources by storing structured information, including textual passages and summaries, organized in a tree format, which enables efficient retrieval and processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional natural language generation systems process full textual passages, then answer quality is maintained, but computational resource consumption increases
Solution Approach 1:
The patent extracts and stores summaries of textual passages in a knowledge graph before querying. When a question is asked, the system retrieves relevant summaries instead of processing complete passages, thereby reducing computational load while preserving essential information needed for answer generation.
Solution Approach 2:
The system performs preliminary summarization of textual passages and stores them in advance in the knowledge graph. This pre-processing step enables faster retrieval and reduces the computational burden during actual query processing, as the summarization work has already been completed beforehand.
2Measurement precision
If comprehensive domain information is stored in the knowledge graph, then answer accuracy improves, but memory consumption increases
Solution Approach 1:
The patent extracts only the essential semantic content from full textual passages and stores summaries in the knowledge graph. This extraction process retains the critical information needed for accurate answer generation while significantly reducing the storage requirements compared to storing complete passages.
Solution Approach 2:
The system stores partial information (summaries) rather than complete passages, which is sufficient for answering most questions. This partial storage approach balances memory consumption with answer accuracy by keeping only the most relevant condensed information.
3Loss of information
If the system retrieves and processes multiple complete textual passages, then information completeness improves, but processing time increases
Solution Approach 1:
The patent retrieves pre-extracted summaries from the knowledge graph instead of complete passages. These summaries contain the essential information needed to answer questions, enabling faster processing while maintaining information completeness for the purpose of answer generation.
Solution Approach 2:
The system performs the time-consuming summarization process in advance and stores the results. During query processing, it only needs to retrieve and compare summaries, which dramatically reduces processing time while preserving the essential information needed for complete answers.
Data Source
AI summary
A natural-language textual sequence representing a question is received from a user. From a knowledge graph domain data is retrieved that includes: a first textual passage based on a first ranking with respect to the natural-language textual sequence, a first textual summary associated with the first textual passage, a second textual passage based on a first ranking with respect to a comparison of the second textual passage and the natural-language textual sequence, a second textual summary associated with the second textual passage, and a third textual summary based on a second ranking with respect to the natural-language textual sequence. An answer to the question from a language model is obtained by providing input to the language model based on the natural-language textual sequence, the first textual passage, the first textual summary, the second textual passage, the second textual summary, and the third textual summary.