Deep Semantic Question Analysis for Knowledgebase Querying
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for converting natural language questions into queries to search databases or knowledgebases are limited by the form of the question and the range of extracted knowledge, and struggle with flexibility across different types of knowledgebases, often relying on predefined queries and surface-level lexicon analysis.
Innovation Solution
A question answering system that employs deep natural language question analysis to create a semantic frame, then generates a question-intermediate expression and converts it into various standardized database queries, allowing for flexible querying across different knowledgebases.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If surface lexicon analysis is used to map natural language questions to predefined queries, then the system is simple and fast, but the system has limitation in question format and knowledge extraction range
Solution Approach 1:
The patent segments the natural language question analysis into multiple hierarchical levels: surface lexicon analysis, deep semantic analysis, and question focus extraction. Each level processes specific aspects of the question independently, allowing the system to handle diverse question formats while maintaining structured processing. The segmentation enables the system to extract question focuses, semantic roles, and relationships without being constrained by surface form.
Solution Approach 2:
The patent transitions from two-dimensional surface lexicon matching to multi-dimensional deep semantic analysis by adding dimensions such as semantic roles, question focuses, entity relationships, and logical structures. This dimensional expansion allows the system to represent and process questions in various formats (who, what, where, when, why, how) uniformly, overcoming the limitations of surface-level analysis.
2Adaptability or versatility
If deep natural language analysis is performed to extract semantic frames and question focuses, then question format flexibility improves, but system complexity increases
Solution Approach 1:
The patent introduces intermediate representation structures (semantic frames, question focus expressions, and standardized query templates) as mediators between natural language input and database queries. These intermediaries bridge the gap between diverse question formats and structured knowledge bases, enabling flexible handling of various question types without requiring complex direct mapping logic for each format.
Solution Approach 2:
The patent changes the parameters of question representation from surface-level lexical features to deep semantic parameters including semantic roles (agent, patient, instrument), question focuses (entity, attribute, relationship), and logical structures. This parameter transformation enables the system to handle format variations by focusing on invariant semantic properties rather than surface form differences.
3Productivity
If predefined standardized queries are used for searching, then search efficiency is high, but the system cannot handle various types of knowledgebases effectively
Solution Approach 1:
The patent creates a universal question processing framework that can handle multiple types of knowledgebases (relational databases, knowledge graphs, document repositories) through a single unified approach. The deep semantic analysis and question focus extraction produce standardized representations that can be mapped to different knowledgebase types, enabling the same system to efficiently query diverse data sources without requiring separate processing pipelines for each type.
Data Source
AI summary
Disclosed are a question answering system for structured knowledgebase using deep natural language question analysis, and a method thereof, the question answering system for structured knowledgebase using deep natural language question analysis includes a deep natural language question analysis unit configured to create a structure of a semantic frame by analyzing a natural language question that is input, a question-intermediate expression creation unit configured to create a question-intermediate expression of a lexicon level based on the semantic frame, a knowledgebase-specialized query creation unit configured to create a query used to search in knowledgebase that is a subject of search, based on the question-intermediate expression, and a knowledgebase search unit configured to find a correct answer in the knowledgebase that is subject of search based on the query, to provide an accuracy of the correct answer, a confidence of the correct answer and an evidence for the correct answer.