Query Normalization for Accurate Database Question Answering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current search engines are unreliable in providing relevant answers to simple questions due to web-based techniques that often return irrelevant results, and users struggle to determine which questions can be answered by the search engine.
Innovation Solution
A method and system for natural language searching that normalizes search queries, matches them with parameterized questions in a database, and provides answers by extracting features and mapping concepts to identify matching database items, allowing for exact or partial matching and using statistical, string distance, or syntactic methods to provide accurate responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If web-based search techniques are used to answer simple questions, then the search engine can provide answers to user questions, but the reliability and relevance of the answers deteriorate due to irrelevant results being returned
Solution Approach 1:
The patent segments the search process into distinct phases: query normalization, feature extraction, database matching, and answer selection. By dividing the search engine functionality into specialized modules (normalization component, feature extraction component, matching component), each can optimize for its specific task, improving overall reliability while maintaining versatility in answering different types of questions
Solution Approach 2:
The patent introduces an intermediary normalization layer between the user's natural language query and the database search. This normalization component acts as a mediator that converts diverse question formats into a standardized form, enabling reliable matching with pre-tagged database questions while maintaining the ability to handle various question types
2Productivity
If query parsing is used to extract relevant terms, then the search engine can process user input, but the ability to accurately answer simple questions deteriorates due to loss of question context and meaning
Solution Approach 1:
The patent applies preliminary normalization to the query before feature extraction and database matching. By pre-processing the query to preserve its semantic structure and context, the system maintains question meaning throughout the processing pipeline while still achieving efficient database querying through the normalized form
Solution Approach 2:
The patent creates a normalized copy of the original query that preserves the essential meaning and context. This normalized version is then used for database matching, allowing the system to process queries efficiently while retaining the semantic information needed to accurately answer simple questions
3Measurement precision
If databases are annotated with answerable questions and parameterized questions are used, then the accuracy of answer matching improves, but the device complexity increases due to database structure and normalization requirements
Solution Approach 1:
The patent transforms questions into parameterized forms with placeholders for variable entities. This parameterization approach maintains high matching accuracy by allowing flexible substitution of specific entities while using a standardized question template structure, thereby managing database complexity through systematic parameterization rather than requiring completely unique question formulations
Data Source
AI summary
Systems and methods for answering user questions using databases annotated with answerable questions are disclosed. The database includes database items organized into columns. Each of the columns may have a question annotated thereto. The question is a parameterized question that is answerable by the database items in the column. A question answering algorithm is used to search the database for an annotated question corresponding to a question that needs to be answered.