Query Intent Mapping for Question Answering Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Question answering systems often misinterpret user intent due to the 'XY Problem,' where users ask indirect questions, leading to incorrect intent determination and retrieval of inappropriate information.
Innovation Solution
A method and system that analyze queries to identify query intents with lower frequency of occurrence and map them to more frequently occurring intents using a frequency distribution map, leveraging associations between rare and common query intents to determine the underlying user intent.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the QA system directly uses the identified first intent for information retrieval, then the retrieval process is simple and fast, but the retrieved information may be inappropriate due to misinterpretation of user intent (XY Problem)
Solution Approach 1:
The patent introduces an intermediary mapping mechanism between the first intent (directly identified from query) and the second intent (mapped from frequency distribution). This mapping structure acts as a mediator that translates the initially identified intent into a more accurate representation of user intent, resolving the XY problem without requiring complete redesign of the intent analysis system
Solution Approach 2:
The system performs preliminary analysis by identifying the first intent before final retrieval, then uses frequency distribution mapping to refine the intent determination. This two-stage approach allows the system to quickly narrow down possibilities first, then apply more sophisticated mapping logic to achieve accurate intent determination
2Reliability
If the system maps rare query intents to common ones using frequency distribution, then intent determination accuracy improves, but the processing time and computational resources increase
Solution Approach 1:
The frequency distribution map is pre-computed and stored before runtime, allowing the system to perform quick lookups during query processing. This preliminary preparation eliminates the need for complex real-time calculations, significantly reducing processing time while maintaining high accuracy
Solution Approach 2:
The system creates a simplified representation (copy) of the intent space through frequency distribution mapping. Instead of performing complex semantic analysis during query processing, it uses the pre-established map to quickly translate queries into appropriate intent categories, dramatically reducing computational overhead
Data Source
AI summary
Embodiments are directed to determining an intent of a query provided to a question answering (QA) system. Embodiments may map a first (e.g., literal) intent of the query to a different, underlying intent. The mapping may be achieved using a map which represents associations between query intents in a first portion of a frequency distribution of known query intents and query intents not in the first portion of the frequency distribution.