UROP Proceedings 2022-23

School of Engineering Department of Computer Science and Engineering 116 Designing Conversational Agents for Neurocognitive Disorders Screening Supervisor: MA, Xiaojuan / CSE Student: LIANG, Danxuan / COMP YANG, Dongjie / DSCT ZENG, Yuhang / DSCT Course: UROP1100, Spring UROP2100, Summer UROP1100, Spring UROP2100, Summer UROP1100, Spring UROP2100, Summer This academic report discusses the potential benefits of using conversational agents for Neurocognitive Disorder (NCD) screening, particularly in healthcare settings. The report presents the Hong Kong Grocery Shopping Dialog Task (HK-GSDT), a simulation game used for NCD screening, and the implementation of a codebook and dialogue flow based on grounded theory to design a conversational agent model for the HKGSDT. The report also outlines the progress made in the project so far, including the completion of essential elements of the codebook, internal testing of the code and dialog flow, and the design of a basic user interface for the wizard’s end. Analysis and Understanding of User Behaviors in Online Communities Supervisor: MA, Xiaojuan / CSE Student: ZHAO, Yubo / DSCT Course: UROP1100, Spring This report presents a quantitative and data-driven approach to gain insights into changes in women's selfawareness. The study employs a bottom-up approach, starting with exploratory data analysis of 2.2k online book comments using the Linguistic Inquiry and Word Count (LIWC) and Bidirectional Encoder Representations from Transformers (BERT) algorithms. The study uses t-tests to identify significant categories related to the research topic. Subsequently, we apply BERT embeddings and use K-means and HDBSCAN clustering algorithms to generate visualizations of the clusters for unsupervised text clustering. Future research will involve training on a larger dataset and establishing a standard for labeling the texts. Ultimately, the goal is to train a reliable model capable of revealing changes in women’s self-awareness through texts. Developing Tools for Privacy-preserving Financial Applications Supervisor: PAPADOPOULOS, Dimitris / CSE Student: GU, Yi / COMP Course: UROP3100, Fall The UROP Program “Developing tools for privacy-preserving financial applications” introduces a decentralized data market application. Based on the summer semester’s study, the author finished the design of the smart contract and its deployment. Now this data marketplace application can communicate with the Ethereum blockchain to ensure the complete operation of smart protocols. At the same time the author designed a separate experiment program to monitor the running time of each part of the whole transaction process. In addition, the author also focused on how to expand the usage of Agora to other research areas and implement a privacy-preserving auction marketplace.