A BERT-based Distractor Generation Scheme with Multi-tasking and Negative Answer Training Strategies.
TL
TL;DR
A new distractor generation scheme with multi-tasking and negative answer training strategies for effectively generating multiple distractors that shows strong distracting power for multiple choice question.
Abstract
In this paper, we investigate the following two limitations for the existing distractor generation (DG) methods. First, the quality of the existing DG methods are still far from practical use. There are still room for DG quality improvement. Second, the existing DG designs are mainly for single distractor generation. However, for practical MCQ preparation, multiple distractors are desired. Aiming at these goals, in this paper, we present a new distractor generation scheme with multi-tasking and negative answer training strategies for effectively generating multiple distractors. The experimental results show that (1) our model advances the state-of-the-art result from 28.65 to 39.81 (BLEU 1 score) and (2) the generated multiple distractors are diverse and shows strong distracting power for multiple choice question.
References24 items
1
Knowledge-Driven Distractor Generation for Cloze-style Multiple Choice Questions
2
Co-Attention Hierarchical Network: Generating Coherent Long Distractors for Reading Comprehension
3
HuggingFace's Transformers: State-of-the-art Natural Language Processing
4
ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
5