Research work i've authored:

Social Behavior of Investors

“Social Behavior of Investors” is my master’s thesis completed on November 2013.

The problem I am trying to solve here is to matchmake investors with companies ( especially early stage startups ) in order to get the investment funds flowing.  I did not focus on public companies as the data are generally public; a great amount of research has been done about them already. But what about the smaller companies where little financial information about them are available ?

The solution is to make model this problem as a classic link prediction problem: given a social network of investors and companies in Time A, predict if (which) new links ( investments ) will form between investors and companies in Time B. I’ve used dataset from Crunchbase for the purposes of this research.

The result is encouraging: the accuracy rate is up to 90% with AUC ( area under curve ) of up to 87% for both datasets[1] that I have used.

[1] I derived 2 datasets from Crunchbase: using Facebook as a seednode, i collected companies, financial organizations and individuals 4 hops away from Facebook. The second dataset is based off RenRen, and collected companies, financial organizations and individuals 4 hops away from RenRen.

Investors Are Social Animals: Predicting Investor Behavior using Social Network Features via Supervised Learning Approach

Published at  ACM: Eleventh Workshop on Mining and Learning with Graphs (MLG 2013), co-located with KDD 2013

Authors: Yuxian, Eugene Liang,  Soe-Tsyr Daphne Yuan

What makes investors tick? In this paper, we explore the possibility that investors invest in companies based on social relationships be it positive or negative, similar or dissimilar. This is largely counter-intuitive compared to past research work. In our research, we find that investors are more likely to invest in a particular company if they have stronger social relationships in terms of closeness, be it direct or indirect. At the same time, if there are too many common neighbors between investors and companies, an investor are less likely to invest in such companies. We use social network features such as those mentioned to build a predictive model based on link prediction in which we attempt to predict investment behavior.

Where's the Money? The Social Behavior of Investors in Facebook's Small World

IEEE/ACM - 2012 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining

Authors: Yuxian, Eugene Liang, Daphne, Yuan Sor-Tsyr

Are investing activities dependent on social relationships? In our research, we apply social network analysis to the field of investing behaviors and discover that investors have a tendency to invest in companies that are socially similar to them. While traditional studies on investing behavior tend to focus on factors like psychology, opinions, investing experience etc, they fail to consider social relationship as an important factor. In this paper we provide general rules of thumb that are useful for companies seeking funding from investor. These rules of thumb are generated by analyzing the social relationships between investors and companies found within the small world of Facebook.


Tools and methods for capturing Twitter data during natural disasters

Published by First Monday ( ) on April 5, 2012

Authors: Axel BrunsYuxian, Eugene Liang

During the course of several natural disasters in recent years, Twitter has been found to play an important role as an additional medium for many-to-many crisis communication. Emergency services are successfully using Twitter to inform the public about current developments, and are increasingly also attempting to source first-hand situational information from Twitter feeds (such as relevant hashtags). The further study of the uses of Twitter during natural disasters relies on the development of flexible and reliable research infrastructure for tracking and analysing Twitter feeds at scale and in close to real time, however. This article outlines two approaches to the development of such infrastructure: one which builds on the readily available open source platform yourTwapperkeeper to provide a low-cost, simple, and basic solution; and one which establishes a more powerful and flexible framework by drawing on highly scaleable, state-of-the-art technology.