UROP Proceedings 2020-21

School of Engineering Department of Computer Science and Engineering 137 Machine Learning Approach to Creating Personalized E-Learning Paths Supervisor: MA Xiaojuan / CSE Student: ZHOU Taichang / COMP Course: UROP1100, Spring This report will introduce a developing software constituted by four major parts: facial feature collection, gaze point detection, engagement analysis, and result exhibition. The facial feature collection is done by OpenFace [1] (an open-source software generating expressive and gestural attributes), which giving out a .csv file. For the gaze point detection, we use data gained from OpenFace to develop a calculate based algorithm. Then we build a Machine Learning algorithm to train our prediction model of students' engagement during e-learning classes. The result exhibition is still under study. Machine Learning Approach to Creating Personalized E-Learning Paths Supervisor: MA Xiaojuan / CSE Student: GUO Bingcan / DSCT Course: UROP2100, Spring Community-based Question Answering forums (CQAs) are important repositories of public knowledge and powerful complement for search engines. How to benefit the sustainable development of the CQA online community has been a popular research topic. Among all the QA forums, academic CQAs are one main category, which are usually dominated by questions about objective fact and requesting professional answers. However, due to the questions’ requirement for answerer’s expertise, many users are hindered from contributing answers and involving in the discussion. Boosting user’s participation in the objective context of academic CQA questions is hence essential. Inspired by the effect of personal pronouns in advertising and relevant research in the subjectivity of CQA questions, we hope to know whether the presence of personal pronouns in academic CQA question titles will benefit the question to encourage users’ active participation in discussion and idea sharing. Our results show that the presence of the personal pronoun ’I’ in a question’s title lead to more answers, while other personal pronouns didn’t show significant improvement in question quality. The personal pronouns are not effective discussion triggers for academic CQA questions. We hope the conclusion can provide preliminary ideas for CQA question quality evaluation and user’s engagement in gamified academic CQA related research.