The NTNU-YZU System in the AESW Shared Task: Automated Evaluation of Scientific Writing Using a Convolutional Neural Network

摘要

This study describes the design of the NTNU-YZU system for the automated evaluation of scientific writing shared task. We employ a convolutional neural network with the Word2Vec/GloVe embedding representation to predict whether a sentence needs language editing. For the Boolean prediction track, our best F-score of 0.6108 ranked second among the ten submissions. Our system also achieved an F-score of 0.7419 for the probabilistic estimation track, ranking fourth among the nine submissions.