Page 64 - Undergraduate Research Opportunities Program Proceedings 2017-18

School of Engineering
Department of Computer Science & Engineering

Supervisor: CHAN Gary Shueng Han / CSE
Student: PAN Linfei / COMP Course: UROP1100, Spring

Recently, the social network analysing has been a popular field for researchers because of its massive potential in
applications. While the social network online has been studied thoroughly, the social network off-lined has not been
well explored. Noticing the strong correlation between the social connection and showing up simultaneously, we use
data mining method to analyze the location information of users generated by the off-line location accumulation facility.
This project puts its main focus on the data generated in a particular mall, and have done some work on building a
model and carrying some experiments.

Supervisor: CHAN Gary Shueng Han / CSE
Student: YAP Alistair Yun Hee / CPEG Course: UROP1100, Fall
UROP2100, Spring

Previous studies on information propagation in location-based social networks (LBSN) such as Yelp and Foursquare make
the simplistic but unrealistic assumption that information propagates exclusively through users ‘actively’ checking-in at
points of interest (POI). This paper introduces a propagation model that accounts for other ‘passive’ means for
information to spread, such as sharing the check-ins of peers via retweets on Twitter or otherwise. The problem of
modelling user-to-user propagation probabilities in LBSN’s is formulated as a classification problem, and initial results
from a variety of classifier models are discussed.

Supervisor: CHAN Gary Shueng Han / CSE
Student: ZHANG Xinyu / COMP Course: UROP1100, Fall
UROP2100, Spring

While there are several categories under recommendation model, so far, only content-based and collaborative-filtering
recommender model have been explored, where the former relies on the entity feature information, and the latter
relied exclusively on affinity data (e.g. rating). Collaborative filtering can be further categorized as Memory-Based
System and Model-Based system. Considering the scalability, we will focus more on the Model-Based system in later
content.

Supervisor: CHAN Gary Shueng Han / CSE
Student: WENG Jiaqi / COGBM Course: UROP1100, Fall
UROP2100, Spring

As I further studied in handling and analytical methods, the collecting process has been refined. The second step after
collecting and reconstructing data in the database is to draft the floor plan. Data collected cannot be used directly
without analyzing and reducing errors. The PDR algorithm is implied to combine all available information and reduce
errors to construct path coordinates. Splitting and merging are applied to the constructed paths to further reduce the
errors. This report will discuss the modifications made on the server to make the receiving process more efficient and
the process applied to the collected data to construct the path and reduce errors. Finally display the analytical results.

54