Ranking Companies Based on Multiple Social Networks Mined from the Web

Many rankings existing for popularity, recommendation, evaluation, election, etc. can be found in the real world as well as on the Web. Many efforts are undertaken by people and companies to improve their popularity, growth, and power, the outcomes of which are all expressed as rankings (designated as target rankings). Are these rankings merely the results of its elements' own attributes? In the theory of social network analysis (SNA), the performance and power (i.e. ranking) of actors are usually interpreted as relations and the relational structures they embedded. For example, if we seek to rank companies by market value, we can extract the social network of the company from the Web and discern, and then subsequently learn, a ranking model based on the social network. Consequently, we can predict the ranking of a new company by mining its relations to other companies. We can learn from existing rankings to expect other target rankings. We can learn from existing rankings to expect other rankings. Furthermore, we can understand the kinds of relations which are important for the target rankings; we can determine the type of structural extension of companies that can improve the target rankings. This study specifically examines the application of a social network that provides an example of advanced utilization of social networks mined from the Web. We present ranking learning approaches using a social network that is mined from the Web. The proposed model combines social network mining and ranking learning, which further uses multiple relations on the Web to explain arbitrary rankings in the real world. Experimental results for learning to rank companies based on multiple social networks mined from the Web confirm the effectiveness of our models for explaining target rankings as well as real world phenomena using multiple social networks. Several findings including social networks vary according to different relational indices or types even though they contain the same list of entities. Relations and networks of different types differently impact on target of ranking. Multiple networks have more information than single networks for explaining target ranking. Well-chosen attribute-based features have good performance for explaining the target ranking. However, by combining proposed network-based features, the prediction results are further improved. This study specifically examines the application of a social network that provides an example of advanced utilization of social networks mined from the Web. 6