Distractor Retrieval Dataset
This benchmark dataset accompanies an article paper titled ``Learning to Reuse Distractors to support Multiple Choice Question Generation in Education''. It contains a test of 298 educational questions covering multiple subjects & languages and a 77K multilingual pool of distractor vocabulary. The goal is for a given question to propose a list of relevant candidate distractors from the pool of distractors.
The dataset is provided in a ZIP file that has two folders. The first folder, 'test-MCQs', contains 6 JSON files, each corresponding to a subject as indicated by its filename namely: English, French, Natural Sciences, History, Biology and Geography. The second folder, `vocab', includes one JSON file containing the list of distractors in a `distractor': `frequency' format.