FAQ System Query Evaluation Using Word Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing FAQ systems face reduced question accuracy due to increased complexity and missing information in queries, often resulting in erroneous answers when supplementing queries with nouns, leading to inadequate responses.
Innovation Solution
A computer-readable medium stores an evaluation program that converts query information into word vectors, evaluates the query based on semantic information associated with these vectors, and performs additional inquiries to gather missing information, ensuring accurate responses by filtering and supplementing queries using a 5W1H perspective.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If the FAQ system performs morphological analysis and supplements queries with nouns to improve query completeness, then the query information becomes more complete, but the question accuracy decreases due to increased complexity and missing information
Solution Approach 1:
The system performs preliminary action by evaluating query completeness using word vectors and semantic information before generating answers. This allows the system to identify missing information in advance and ask targeted supplementary questions, rather than attempting to supplement queries with nouns that may reduce accuracy. The evaluation process checks whether the query contains sufficient information to provide an accurate answer, and only then proceeds to answer generation.
2Measurement precision
If the system asks further questions to supplement missing information, then the query accuracy improves, but the response time increases due to additional interaction steps
Solution Approach 1:
The system applies partial action by selectively asking supplementary questions only for the specific elements that are missing or insufficient in the query, rather than asking all possible questions. The word vector evaluation identifies which semantic elements are weak or missing, and the system asks targeted questions only for those specific elements, thereby reducing the number of interaction steps while maintaining query accuracy.
3Productivity
If the FAQ system operates outside staffed business hours, then the service availability improves, but the ability to handle complex queries with missing information deteriorates
Solution Approach 1:
The system implements self-service by automatically evaluating query completeness using word vectors and semantic information without human intervention. The system independently determines whether a query contains sufficient information, identifies missing elements, and asks supplementary questions autonomously. This allows the FAQ system to operate effectively outside staffed hours while maintaining query evaluation accuracy through automated semantic analysis.
Data Source
AI summary
The information processing device receives question information, converts words contained in the received question information into word vectors including a plurality of element values respectively associated with a plurality of elements. The information processing device refers to the memory that stores, in association with the elements, specific information specifying the strength of specific meanings contained within the element values associated with the elements and acquires, using the specific information, a plurality of semantic information items indicating the strength of the specific meanings respectively contained within the plurality of element values included in the word vectors. The information processing device performs an evaluation relating to the received question information on the basis of the acquired plurality of semantic information items.