Advisor: QU Huamin/CSE
Student: CAO Wang/ELEC-HR UROP Course: UROP1100, Fall 2014 For potential postgraduate students who are preparing to apply for a certain program, the official information from the university indicating the graduation intervals, the number of students that a particular professor typically supervises per year, or the trend of research areas that have attracted increasingly more participants usually is of great assistance in their selection of a suitable postgraduate program. For the Department of Computer Science at HKUST, however, the archives are arranged discretely on different Web pages in traditional text form, which is inconvenient for visitors wanting to acquire information on past circumstances. Therefore, we want to develop a data visualization approach that intuitively presents the general patterns of postgraduate computer science programs at HKUST.
Advisor: QU Huamin/CSE
Student: JEONG Seowon/MATH-ST UROP Course: UROP1100, Fall 2014 This report traces the developmental trends of HKUST and KAIST, two leading Asian USTs, since their inception. The multidimensional approach was employed to scrutinize the similarities and differences in their development. Although the fundamental mission of HKUST and KAIST appears to converge, because of the interference of their cultures and geographic locations, certain aspects such as student interest exhibit distinct heterogeneity. Absolute superiority is unquantifiable because both universities have relative strengths and weaknesses; whereas HKUST is more successful in extending international partnerships for trading intellectual property, KAIST is more actively engaged in research. These institutions’ assignment of different weights to the same domain has changed their developmental orientation continuously.
UROP Proceedings 2014-15 Advisor: QU Huamin/CSE
Student: LO Tsz Cheung/COMP UROP Course: UROP1100, Fall 2014 The purpose of this project was to visualize information on qualifying exams, proposal defenses, and thesis defenses from students who have finished or were still active in their postgraduate programs in the Department of Computer Science at HKUST, from 2000 to 2014. The entire project was divided into two parts, which included using a Web crawler to extract extant data from official archives and performing visualization for a clear reference.
Keywords: postgraduate program, information visualization.
Advisor: QU Huamin/CSE
Student: ZHANG Yifeng/COMP UROP Course: UROP1100, Fall 2014 The number of commercialized software programs is increasing at an unprecedented rate in the twenty-first century. However, with the proliferation of the software industry, source code plagiarism prevention and detection are becoming increasingly crucial for ensuring the healthy development of the field. Among all of the properties of programs, the coding style is an essential criterion for determining whether plagiarism has occurred. The project presents several data visualization methods for analyzing the coding style, which could reveal the level of similarities between two programs with superior results compared with extant numerical tools. Moreover, examples of diverse methods are provided for clarification.
Keywords: software programs, plagiarism detection, coding style, visualization.
Advisor: QU Huamin/CSE Co-advisor: CHEN Lei/CSE
Student: CHEN Taiyou/COMP UROP Course: UROP1100, Summer 2015 That selecting and pursuing a major or career that suits a student’s interests and passions is critical is unanimously supported. Nevertheless, choosing an ideal major remains a critical challenge for students and their parents. A considerable amount of research has been conducted for assisting people identify the majors that suit them, and how their major fits them. Specifically, ACT, Inc. has collected multiple factors including student indications and declarations of their major, ACT test scores, program of study, and so on, to build “best-fit” major and “fitting indices” for every student.
However, the performance of ACT assessments—how the perceived major suits the students’ actual interests—is still to be examined. This study emphasizes mainly the analysis and visualization of the related factors, and aims to determine whether the perceived interest suits the actual major.
Keywords: visual analysis, fitting index, change of interest.
UROP Proceedings 2014-15 Advisor: QU Huamin/CSE Co-advisor: CHEN Lei/CSE
Student: CUI Minhui/SSCI UROP Course: UROP1000, Summer 2015 At present, students of higher education tend to have a preference for U.S. universities, which have accumulated major data sets on admissions information. Therefore, research on the regularity of admission outcomes is required to obtain sufficient evidence for this large group with this specific interest. People may exercise their critical thinking abilities, and substantial research has been conducted to identify the trends. However, no good visualization has been obtained, and it is difficult to determine the regularity of universities directly. This paper presents a conclusion based on selected top U.S. universities by using two visualization methods, and provides the results.
Keywords: top universities, admissions outcomes, big-data visualization.
Advisor: QU Huamin/CSE Co-advisor: CHEN Lei/CSE
Student: DU Xinnan/SENG UROP Course: UROP1000, Summer 2015 With the development of new technologies, it is becoming increasingly difficult for individual researchers to conduct groundbreaking research on their own, and thus, scientific collaboration has become increasingly critical. Interest has recently been growing in exploring and visualizing collaborations among various entities. As one of the best research universities worldwide, many research projects are conducted each year at HKUST. However, scientific collaboration has not been fully analyzed. The objective of this project is to provide a general overview and present detailed information on the evolution of collaboration by using publication data provided by the HKUST library. By using various visualization techniques, we identified several patterns across schools, departments, and professors. Specifically, we used node-link diagrams with a force-directed layout [3] to determine the general trend of collaborations directly, and more detailed information is provided through line charts and area charts [5].
Advisor: QU Huamin/CSE Co-advisor: CHEN Lei/CSE
Student: PURI Abishek/MAEC UROP Course: UROP1000, Summer 2015 This paper presents a discussion on the visual analytic system ParkVis, which our team developed for the IEEE VAST Challenge 2015. The data set the system was built for amusement park data, specifically the fictitious DinoFun World Park, which was used for the challenge. This system allows users to filter movement data of parks visitors based on which areas they visited and the rides they went on, and it also scales all of the movement data of a user by employing our MDS and presenting it in MDS view. In this report, I focus primarily on the visual aspects of our system as well as the interactions between various components in the system.
UROP Proceedings 2014-15
Advisor: QU Huamin/CSE Co-advisor: CHEN Lei/CSE
Student: WANG Tianyu/MATH-AM UROP Course: UROP1100, Summer 2015 This study investigates the role that force-directed graph can potentially play in dimension reduction, which is a category of algorithms that embed points in a high-dimensional space into a low-dimensional space through projection or transformation. To achieve satisfactory results in dimension reduction for visualization through force-directed graphs, we first constructed a graph from high-dimensional data. Afterward, we attempted to discern a suitable layout by fixing certain points in the graph before the points began to evolve with gravity. This report details our attempt to solve the dimension reduction problem from a new perspective.