Keyword Extraction Using Latent Semantic Analysis For Question Generation

G. Deena ； K. Raja

《淡江理工學刊》 26卷4期 (2023/04) Pp. 501-510

https://doi.org/10.6180/jase.202304_26(4).0006

摘要

In-Text Mining, Information Retrieval (IR), and Natural Language Processing (NLP) dig out the important text or word from an unstructured document is coined by the technique called Keyword extraction. It helps to identify the core information about the document in specific. Instead of going through the entire document, this method helps to retrieve sufficient information instantly in a short span of time. It is essential to mine the meaningful word from the document in text analytics. The proposed system has been based on semantic relation to extracts the keyword from unstructured text documents by means of practice like Latent Semantic Analysis (LSA). In view of this method, there exists a semantic relation between the sentences available in the document and the words. Extracted text permits to signify text in a strong way and has a high preference to carry more important information about the sentences. In this regard, LSA has produced better outcomes when compared with the TF-IDF, RAKE, YAKE, and Text Rank algorithm. Consequently, the keyword extraction has been occupied in Automatic Question Generation (ACQ) to generate the Fill up the blank (FB) and Multiple Choice Questions (MCQ) with distractor set. The top five, ten keywords are involved in questionable generation. The proposed system could be implemented in the question generation system to assess the skill level of the learner.

關鍵字

Natural Language Processing ； Latent Semantic Analysis ； Multiple Choice Questions ； Keyword Extraction ； Semantic relation