KNIGHT: Knowledge Graph-Driven Multiple-Choice Question Generation with Adaptive Hardness Calibration

Mohammad Amanlou, Erfan Shafiee Moghaddam, Yasaman Amou Jafari, Mahdi Noori, Farhan Farsi, Behnam Bahrak · 2026

Read KNIGHT: Knowledge Graph-Driven Multiple-Choice Question Generation with Adaptive Hardness Calibration by Mohammad Amanlou, Erfan Shafiee Moghaddam, Yasaman Amou Jafari, Mahdi Noori, Farhan Farsi, Behnam Bahrak (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-20135.txt.

Description

Good test datasets for language models take time and money to build. That bottleneck slows testing of other applications. A 2026 arXiv paper by Mohammad Amanlou and five co-authors introduces KNIGHT, a framework that generates question datasets from external sources using a knowledge graph.

KNIGHT first builds a topic-specific knowledge graph. That graph is a structured, compact summary of items and relations. Once built, it can be reused to set difficulty levels under teacher control, including multi-hop questions, without re-feeding the full source text each time. The graph acts as a compressed, reused state, so generating questions becomes a cheap read over the graph. The framework is shown on Wikipedia and Wikidata, while staying free of any single domain or ontology. As a case study it produces six question datasets in history, biology, and mathematics, judged on fluency, clarity, topic fit, option uniqueness, and more.

Catalog record

CREATORMohammad Amanlou, Erfan Shafiee Moghaddam, Yasaman Amou Jafari, Mahdi Noori, Farhan Farsi, Behnam Bahrak
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE95 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-20135

Browse the full library · About the Tower of Babel