Summary of - A novel approach to generate distractors for Multiple Choice Questions Authors

Document Type

Article

Abstract

Study Background: The research presented titled “A novel approach to generate distractors for Multiple Choice Questions” by Dr. Archana Praveen Kumar, Dr. Ashalatha Nayak andDr. Manjula Shenoy is centered on the creation incorrect options or distractors for Multiple Choice Questions (MCQs). The approach aims in generating distractors in order to evaluated the cognitive skills of the examinees.

Research Goals and Hypotheses: The primary goal is to obtain distractors for the Multiple Choice Questions based on semantics and context of the question using Bidirectional Encoder Representations from Transformers (BERT).

Methodological Approach:

  1. Generates MCQ distractors for a technical domain based on semantics and context using BERT, Ontology and WordNet.
  2. Generates distractors using both structured and unstructured data.
  3. Evaluates distractors based on Item Response Theory i.e. Question Difficulty Level, Distractor efficiency and their discrimination.

Results and Discoveries:

  1. Among 114 MCQs the proposed approach DIstractor GENeration (DIGEN) was able to generate 97 questions with four distractors, 16 questions with three distractors and 1 question with one distractor.
  2. DIGEN has been able to generate nine ‘Most Difficult’, five ‘Difficult’, thirty-seven ‘Normal’, twenty-two ‘Easy’ and forty-one ‘Most Easy’ distractors.
  3. Generation of distractors for 114 questions from DIGEN showed that 99 questions could be used for assessment based on the distractor efficiency and discrimination.
  4. The distractors from DIGEN also when evaluated for question difficulty level showed that 73 from 114 questions were very good questions to be used for assessment based on Bloom’s Taxonomy.

Publication Date

2023

Recommended Citation

Kumar, A. P., Nayak, A., Shenoy, M., & Goyal, S. (2023). A novel approach to generate distractors for multiple choice questions. Expert Systems with Applications, 225, 120022.

Publication Date

2023

Link to Full Text

Share

COinS