자연어 툴킷 및 AI 챗봇을 활용한 초등영어 어휘평가 자동화 알고리즘 개발 연구

A study on the development of an automated algorithm using natural language toolkit (NLTK) and artificial intelligence (AI) chatbot for primary English vocabulary assessment

The purpose of this study aims to develop the algorithm that automatically assesses English vocabulary from students’ discourses by using the natural language toolkit (NLTK) and the artificial intelligence (AI) chatbot. The way to build the algorithm was as follows: First, three task-based AI chatbots were built by using Google Dialogflow API (Application Programming Interface) and discourses were transcribed automatically by the API’s history function. Second, the vocabulary data from three objective assessment criteria (Compleat Lexical Tutor VP-kids, CEFR, CEFR-J) were compared with textbooks, and new criteria were reorganized. Third, the discourses were tokenized and the parts of speech (POS) of the words utilized throughout the discourses were tagged by using Python programming language and the NLTK. Furthermore, different meanings of the vocabularies, depending on contexts, were analyzed and graded through Python text mining. Finally, the results were confirmed by experimentally distributing the utterance data from a 6th-grade student interaction with a chatbot among the three chatbots used in the study. Through the application of the algorithm, the vocabularies were evaluated properly by showing the expected results. However, the gradual development of the NLTK POS tagger is needed as some words were tagged incorrectly. Nevertheless, this algorithm showed the possibility of a collaboration of human, AI chatbots, and automation technology for highly efficient English assessment.