Designing, Analyzing and Exploiting Stake-based Social Networks Tsung-Ting Kuo*, Jung-Jung Yeh, Chia-Jen Lin, Shou-De Lin Graduate Institute of Networking and Multimedia, National Taiwan University, Taipei, Taiwan *d97944007@csie.ntu.edu.tw Abstract—It is widely recognized that stakeholder information can provide important knowledge about stock investments, and an increasing number of countries require that such information is publicly available. In this paper, we present a novel way to exploit stakeholder information by using it to construct stake- based social networks, namely, StakeNet. We also provide a visualization tool that displays socio-centric and ego-centric views of the networks. In addition, we analyze stakeholders’ static and dynamic behavior patterns in StakeNet, and demonstrate that most of StakeNet’s properties are similar to those of a typical social network, except that the in-degree distribution does not follow a power law distribution. Finally, we demonstrate two applications of StakeNet by exploiting it to identify important companies and to group companies together. The experiments show that our results are highly consistent with the outcomes generated by human experts. Source code, dataset, and resources are available at http://www.csie.ntu.edu.tw/~d97944007/stakenet/ Keywords-social network analysis; stakeholder analysis; stakeholder management I. INTRODUCTION The term “stakeholder” can be defined as “any group or individual who can affect or is affected by the achievement of the firm’s objectives” [1]. In the information era, data about the stakeholders of publicly traded companies is being made available to investors in an increasing number of countries. Stakeholder data is important to stock investors because it provides information about individual companies and the relationships between companies. However, comprehending and utilizing stakeholder data is difficult because of the enormous volume available and the amount of detail involved. Furthermore, stakeholder data changes over time, and its dynamic nature makes the interpretation and usage of the data even more difficult. As a result, an intuitive and effective way to present, analyze and exploit stakeholder data is highly desirable. In this paper, we propose a social network called StakeNet, which is constructed using stakeholder data. StakeNet is a directed, weighted, dynamic, and heterogeneous social network. We also investigate three issues for StakeNet. First, we provide a visualization tool that enables investors to view the relationships among companies and stakeholders in an in-depth and efficient manner. The tool can be utilized to examine the relationships of any given company as well as the overall market environment. Second, we represent Taiwan’s stock market using StakeNet and perform on top of it both static and dynamic social network measures, such as the degree distribution, clustering coefficient, giant connected component analysis. Finally, we demonstrate the value of StakeNet by using it in two applications: rank important companies, and group companies into intra-related groups. The experiment results show that our system can achieve very high consistency comparing to the results generated by experts in investment companies. II. STAKENET CONSTRUCTION AND APPLICATIONS A. Constructing StakeNet We define StakeNet as a graph SN t-t’ = {V, E}, where V = {v 1 , v 2 , …, v n } is a vertex set, and E = {e ij = edge from v i to v j | 1 ≤ i, j ≤ n, i ≠ j} is an edge set. For each vertex v k , type(v k ) ∈ {person, company}; and for each edge e ij , type(e ij ) ∈ {hold, manage, transfer}. Weight(e ij ) equals the market value of the stocks (i.e., shares multiplied by prices) held or transferred by edge e ij if type(e ij ) = hold or transfer. Note that the weight is zero if type(e ij ) = manage. Each type of edge can only occur between certain types of vertex, as shown in Table 1. For each specific time point, there is a corresponding StakeNet, since relationships can change over time. We conducted a survey of the stakeholder information available in seven stock markets. The results show that Japan and Taiwan provide the most comprehensive publicly available stakeholder information. As a result, we took the Taiwan stock market as the data source and gathered stakeholder information from the official website of the Taiwan Stock Exchange 1 . The data covered the period 2002/10 to 2009/10, a total of 85 months. There were 2,026 publicly traded companies registered on the Taiwan stock market during that period. TABLE I. EDGE AND VERTEX TYPES IN STAKENET. Type of Edge eij Type of Vertex From vi To vj Hold Person or Company Company Manage Person Company Transfer Person Person B. Interrelation Visualization We have developed a visualization tool that provides two views of a network: a socio-centric (global) view and an ego- centric (local) view. The socio-centric view displays the whole network, and the user can specify a minimum degree to control which vertices are displayed. In the ego-centric view, the user can specify a specific person / company and the level of interested neighbors, and the system will display a subgraph centered at the person /company for further investigation. The socio-centric view (i.e., the global view of all nodes) in StakeNet constructed from 2008/11 to 2009/10 is shown in We acknowledge the CyberLink Corporation (http://www.cyberlink.com ) for financial support of this work. 1. Available at http://www.twse.com.tw 2010 International Conference on Advances in Social Networks Analysis and Mining 978-0-7695-4138-9/10 $26.00 © 2010 IEEE DOI 10.1109/ASONAM.2010.14 402 2010 International Conference on Advances in Social Networks Analysis and Mining 978-0-7695-4138-9/10 $26.00 © 2010 IEEE DOI 10.1109/ASONAM.2010.14 402 2010 International Conference on Advances in Social Networks Analysis and Mining 978-0-7695-4138-9/10 $26.00 © 2010 IEEE DOI 10.1109/ASONAM.2010.14 402