Search

A huge and growing number of scientific papers are authored by non-native English speakers, driving increased demand for effective computer-based writing tools to help writers composing scientific articles. The Automated Evaluation of Scientific Writing (AESW) shared task promotes the use of natural language processing tools to improve the quality of scientific writing in English by predicting whether a given sentence needs language editing or not. In this study, we propose an Ensemble Multi-Channel Neural Networks (called EMC-NN) model for scientific language editing evaluation, comprised of three main parts: a multi-channel word embedding representation, a combination of Bidirectional Long Short-Term Memory and Convolutional Neural Networks, and a majority voting ensemble. Experimental results on 143,804 testing sentences show that our proposed EMC-NN achieved an F1-score of 0.6367, outperforming the winner of the AESW-2016 competition task and the recent BERT transformers. Based on a series of in- depth analyses comparing the number of channels, ensemble size and network architectures, the proposed EMC-NN model is a relatively simple, but effective approach that offers significant performance improvements for scientific writing evaluation tasks.

Recently, the literature reported that multimedia learning is more appealing and can increase learning effectiveness. Furthermore, cued retrospective reports (CRRs) can enhance students' learning performance, especially for the novices lacking prior knowledge. In this work, a multimedia learning assistant system is constructed by applying an eye-tracking technique to record and evaluate learner's CRRs. First, each student takes the test after watching the multimedia teaching animations. Those who cannot pass the test will be provided to watch the same video but with segmented video clips to reduce possible high cognitive load. Notably, one-class SVM classifier was applied to automatically judge whether the learners' concepts are correct. Experts' or high achievement peers' CRRs are provided to assist the low-achievement learners. Then, those low-achievement learners accept the second test to verify the effectiveness of our proposed system. 31 students in a junior high school participated in the pre-test, and 16 students did not pass the first test after watching the video. The experimental results revealed that these 16 students have increased their grades from 53.75 to 75. In addition, the classification rate can reach up to 94.59%. The proposed learning assistant system can effectively assist teachers in providing realtime assessment and feedback.

Structured argumentation support environments have been built and used in scientific discourse in the literature. These environments emphasized either on exchanging information or on constructing arguments for presentation. In this work, an intelligent online asynchronous argumentation platform that detects whether the learners address the expected discussion issues is proposed. After each learner issues an argument on the learning platform, a term weighting method is adopted to derive inputs parameters of a SVM classifier. The classifier then determines if the learners' arguments are related to the concept maps outlined by the instructor. Notably, a peer review mechanism is established in this work to improve the quality of term weighting approach. The experimental results revealed that the students in a junior high school participating in argumentation learning activities of natural science were benefited by the proposed argumentation based learning assistance platform.