Diverse Question-Answer Pair Generation via Query-Focused Summarization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing question-answer (QA) pair generation technologies lack diversity in question types and quality, which is crucial for comprehensive learning and assessment in educational settings.
Innovation Solution
A device and method incorporating a query-focused summarization (QFS)-based answer generator, an iterative QA generator, and a relevancy-aware ranker to generate diverse QA pairs, ensuring high-quality and relevant outputs by expanding candidate sets and prioritizing top-ranked pairs based on relevance and diversity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional question-answer pair generation methods are used, then the generation process is simple, but the diversity of question types and quality are insufficient
Solution Approach 1:
The generation system is divided into three independent modules: query-focused summarization module, answer generation module, and question generation module. Each module performs a specific function, allowing the system to generate diverse QA pairs by coordinating these segmented components rather than using a monolithic complex system.
Solution Approach 2:
The query-focused summarization is performed in advance before answer and question generation. This preliminary action extracts key information from the passage, providing a structured foundation that guides subsequent answer and question generation, thereby improving diversity without requiring complex real-time processing.
2Adaptability or versatility
If more question-answer pair candidates are generated to increase diversity, then QA type diversity improves, but the quality and relevance of individual pairs may deteriorate
Solution Approach 1:
The system incorporates a relevancy-aware ranker that provides feedback on the quality and relevance of generated QA pairs. The ranker evaluates generated pairs against the original passage and query, using this feedback to select the most relevant pairs while maintaining diversity. This feedback mechanism ensures quality without sacrificing the expansion of candidate numbers.
Solution Approach 2:
The system dynamically adjusts generation parameters such as the number of candidates to generate and the ranking thresholds based on the complexity of the passage and query. By changing these parameters adaptively, the system can generate sufficient diverse candidates while maintaining quality through parameter optimization rather than fixed rigid rules.
Data Source
AI summary
Disclosed is a device and method for educational question-answer pair generation (QAG) considering type diversity. The method for question-answer pair generation is performed by a computing device and includes generating a query-focused summarization (QFS) for a passage; generating an initial answer based on the passage and the QFS; generating a question corresponding to the initial answer based on the initial answer, the passage, and an interrogative word; generating an answer corresponding to the question based on the question and the passage and generating a question-answer (QA) pair; and deriving a final QA pair by selecting at least one QA pair from among the QA pairs.