UROP Proceedings 2020-21

School of Engineering Department of Computer Science and Engineering 140 Predicting Student Performance on an E-Learning Platform Supervisor: MA Xiaojuan / CSE Student: ZHANG Yuanhao / COMP Course: UROP3100, Summer Nowadays, gamification, the use of game design elements in non-game contexts, is a widely used measure for some learning applications (e.g., Duolingo) to stimulate users’ interest and passion for learning and to increase their learning efficiency. Nevertheless, gamification is not always successful as it sometimes distracts users from their initial goal, learning. To get the bottom of this phenomenon, we conduct qualitative research by using Duolingo as a case study. With an extensive dataset generated from the Duolingo forum over the last nine years, we run a content analysis and summarize the main ramifications of such misusing behavior and the reasons behind it. Moreover, we try to decipher them and make some reasonable suggestions to resolve or undercut these issues. A Study on the Principles and Implementations of Array Databases Supervisor: NG Wilfred Siu Hung / CSE Student: HAU Yiu Tong / COMP Course: UROP1100, Fall Data Versioning in database systems is a crucial feature for conducting scientific research. As a popular choice for large-scale data management in scientific communities, distributed array database systems require an effective storage mechanism for data versioning. In this report, we present a method for finding an optimal storage strategy for distributed array databases. First, the problem is formulated into a graph problem, then using a multilevel k-way graph partitioning scheme, version arrays are separated into different partitions representing computing clusters where similar arrays are stored together. Then, each partition is optimized to find an effective storage solution for storage space and array recreation time. Develop a Machine Learning Application on Video Data Base Retrieval Using Tensorflow Facilities Supervisor: NG Wilfred Siu Hung / CSE Student: TSE Wai Chung / SENG Course: UROP1100, Summer Semantic video retrieval is one of the most common requirements for computer users. That is, given a user query, we want to extract all the videos that are semantically related to the query, without requiring the user to do any preliminary tagging on the videos. This report presents an approach to tackle this task. With the support of TensorFlow, we applied transfer learning to state-of-the-art convolutional neural networks Xception and I3D Kinetics, allowing them to interpret videos and automatically assign them with tags. Then, word embedding semantic search techniques are applied to multi-match user queries to auto-tagged videos, achieving the task of semantic video retrieval.