Scholarlink

<p>The most crucial aspect of any teaching–learning process is evaluation and assessment. Assessment may be formative or summative in nature. For summative assessment, mostly a question-based evaluation is followed. The assessment questions can be used to identify a learner's abilities and learning outcomes. Hence, creating a variety of questions is an open-ended problem and requires laborious efforts from domain experts. This problem of question creation can be automated with the help of automatic question generation (AQG). AQG aims to generate questions from a piece of natural language text. The paper put forth a novel model for AQG for Marathi language text. In this research paper, a working model for AQG in the context of postpositional noun phrases in Marathi text has been presented. As part of our working philosophy, a regular expression-based shallow parser for extracting noun and postpositional phrases from the Marathi sentence has been utilized. A rule-based tree grafting method has been used to generate an interrogative sentence from the parsed sentence. The empirical evaluation of the proposed methodology has been carried out with the help of manual evaluations, yielding an accuracy of 87.25% for the adequacy level metric, 87% for the fluency metric, and 58% for the difficulty level metric on the corpus chosen from the fifth and sixth grade science text book.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Automatic Question Generation from Postpositional Phrases of Marathi

  • 期刊论文

  • 发表日期：

    2025年2月19日

查看全文
  • Pushpa M. Patil,
  • R. P. Bhavsar,
  • B. V. Pawar

摘要

The most crucial aspect of any teaching–learning process is evaluation and assessment. Assessment may be formative or summative in nature. For summative assessment, mostly a question-based evaluation is followed. The assessment questions can be used to identify a learner's abilities and learning outcomes. Hence, creating a variety of questions is an open-ended problem and requires laborious efforts from domain experts. This problem of question creation can be automated with the help of automatic question generation (AQG). AQG aims to generate questions from a piece of natural language text. The paper put forth a novel model for AQG for Marathi language text. In this research paper, a working model for AQG in the context of postpositional noun phrases in Marathi text has been presented. As part of our working philosophy, a regular expression-based shallow parser for extracting noun and postpositional phrases from the Marathi sentence has been utilized. A rule-based tree grafting method has been used to generate an interrogative sentence from the parsed sentence. The empirical evaluation of the proposed methodology has been carried out with the help of manual evaluations, yielding an accuracy of 87.25% for the adequacy level metric, 87% for the fluency metric, and 58% for the difficulty level metric on the corpus chosen from the fifth and sixth grade science text book.