• Journal of Digital Convergence
• /
• v.19 no.9
• /
• pp.1-8
• /
• 2021

The purpose of this study is to identify the influence of users' satisfaction with the Automatic Writing Evaluation(AWE) on learners' sense of learning achievement through self efficacy in English writing class. AWE is a tool that automatically provides feedback on writing outputs by AI technology. College students were asked to write essays for each topic and use AWE to get feedback on their drafts, and finally revise them referring to the feedback. A questionnaire survey was conducted for the data collection. The data was analyzed using SPSS, and smart PLS-SEM along with bootstrapping techniques, The results of the study reveal the followings: 1) the convenience and usefulness of AWE had a positive effect on the willingness to reuse it; 2) the satisfaction with AWE had a positive effect on self-efficacy; 3) self-efficacy had a positive effect on learning achievement in terms of emotional and linguistic aspects. With the development of the 4th industry and A.I. technology, it is recommended to introduce new materials or programs such as AWE in English education.

• Journal of KIISE
• /
• v.42 no.9
• /
• pp.1124-1132
• /
• 2015

In order to train a supervised machine learning algorithm, it is necessary to have non-biased labels and a sufficient amount of training data. However, it is difficult to collect the required non-biased labels and a sufficient amount of training data to develop an automatic English Composition scoring system. In addition, an English writing assessment is carried out using a multi-faceted evaluation of the overall level of the answer. Therefore, it is difficult to choose an appropriate machine learning algorithm for such work. In this paper, we show that it is possible to alleviate these problems through ensemble learning. The results of the experiment indicate that the ensemble technique exhibited an overall performance that was better than that of other algorithms.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.04a
• /
• pp.787-790
• /
• 2015

본 연구에서는 글의 구성요소를 추측하는 가장 높은 성능을 나타내는 알고리즘을 제시한다. 실험 방법은 글의 각 문장에 대한 자질을 추출, 자질 선택, 그리고 데이터에 대해 여러 기계학습 알고리즘을 학습시킨 후 성능을 비교하여 진행하였다. 또한 이 중 가장 높은 성능을 보이는 CRF를 기존에 연구되어 있는 성능과도 비교하였다. 마지막으로 CRF가 구성요소를 추측하는 데 있어서 가장 높은 성능을 보이는 이유에 대해 분석하였다. 국내의 유명 어학원 및 토플 웹사이트를 통해 1969개의 토플 에세이를 수집했으며 2명의 전문 평가자를 통해 각 문장을 8개의 분류로 나누었다. 이를 CRF를 적용한 결과 87.2%의 F score가 나왔으며 기존 연구결과, 그리고 다른 알고리즘보다 높은 성능을 보였다.