Multiple-Choice Item Generator for Narrative and Declarative Texts

Abstract

Abstract-Multiple Choice Item Generator is a research focused on generating grammatically and semantically correct questions and generating plausible distractors. The system's framework has three main modules: pre-processing, question generation and distractor selection. Pre-processing module implements Hobbs Algorithm for anaphora resolution and direct to indirect speech conversion for handling quoted statements. In generating questions, question overgeneration and ranking framework of Heilman and Smith is used. The distractor selection module uses collocation extraction and Wordnet-based methods such as Lin's semantic similarity measure, hypernyms, hyponyms and maximal bipartite graph matching .Index Terms-Hobbs algorithm, question overgeneration, multiple-choice item generator.