978-1-61284-952-2/11/$26.00 ©2011 IEEE 1069 IEEE Int'l Technology Management Conference Finding Business Partners and Building Reciprocal Relationships – A Machine Learning Approach Junichiro Mori, Yuya Kajikawa, and Hisashi Kashima The University of Tokyo 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-8656 JAPAN {jmori,kaji}@ipr-ctr.t.u-tokyo.ac.jp, kashima@mist.i.t.u-tokyo.ac.jp Abstract Business development is vital for any firms. However, globalization and the rapid development of technologies have made it difficult to find appropriate business partners such as suppliers, customers and outsources and build reciprocal relationships among them, while it simultaneously offers many opportunities. In this contribution, we propose a new computational approach to find business partner candidates based on firm profiles and transactional relationships among them. We employ machine learning techniques to build prediction models of customer-supplier relationships. We applied our approach to Japanese firms and compared our prediction results with the actual business data. The results showed that our approach successfully found plausible candidates and reciprocity among them whose accuracy is about 80%. Using machine learning approach, we have the accuracy of predicting a customer-supplier relation of 84%, and the accuracy of predicting a reciprocal customer-supplier relation is about 75-79%. These results show that our approach can be a new powerful tool to develop one's own business in the complicated, specialized and rapidly changing business environments of recent years. Introduction Business development is a perdurable issue for any firms pursuing sustainable development. It consists of a series of activities including assessment of market opportunities, exploration of plausible business partners, business model design, formal proposal, and follow-up sales activities. In addition to direct effect increasing turnover by customer acquisition, organizational relationships among firms such as customer-supplier relationships and strategic alliance have an indirect effect to develop business because they can work as a source of innovation. Firms in a cooperation network can utilize their network in a variety of ways. They not only share the costs and risks of their activities but also obtain access to new markets and technologies, make use of complementary skills, and share knowledge capabilities, and finally influence firm performance [10, 11, 18, 24, 25, 31]. To exploit the opportunities acquired by partnerships, firm must seek not only for new business partners but also reciprocal business models assuring long-term relationships. Business reciprocity is the term to describe business dealings between independent firms whereby they make mutual concessions designed to promote the business interests of each [28]. Although researchers have studied the origin of business reciprocity from a point of buying power like buying power and profitability of customer [16, 22], the current business situation is more complex and the understanding on the reciprocity is not enough. [15] argued that reciprocity comes from the cross-managerial level between firms, however, the origin of such a managerial strategy and characteristics of those firms building reciprocal relationships are obscure. In daily business expertise, much effort is devoted to gain new customers and to penetrate existing markets. Firms must seek new business partners to acquire new opportunities and also to activate existing relationships, while long-term and dedicated relationships with trust is the vital source of innovation. But such an activity is laborious, time-consuming, and subjective, while the performance is the critical factor of the success of business. Traditional stakeholder theories and frameworks can enhance our understandings in the roles of business partners, but cannot serve as a guidance to find new partners and to keep reciprocity among them. And current expertise employ a manual approach to analysis and do not scale up to accommodate the rapid pace of change in business environment and market. While there are several papers to evaluate selected firms by financial strength, less works exists to find plausible business partners, especially, customer and supplier, among a vast amount of firms. In the previous literatures, attributional features including financial data, product and service quality, technological capabilities, compatibility, and etc. are included in the analysis to evaluate suppliers [12, 13, 30, 33] by introducing the implications of previous literatures in supply chain management [5, 6, 8, 9, 20, 23, 29, 34]. Although supplier candidates are usually evaluated by quality, cost, delivery, it is hard to know them in advance before actual transactions is contracted and launched. Therefore, in this paper, we use explicit data of firms like the number of employee, date of foundation, which can be known before starting business with plausible business partners. The aim of this paper is to model and predict the existence of reciprocal customer-supplier relationships. Therefore, the issue tackled in this paper is closely related to the link-prediction problem. In the link-prediction problem, utilization of relational features intrinsic to the network itself can offer meaningful inferences from observed network data [21]. We utilize the customer supplier relationships among the other firms and with other firms to predict the focal relationship. There are fewer studies utilizing such relational features in business data mining. Additionally, previous works typically use a small corpus whose size is around 100. In order to adopt such an approach, we use the large size whose size is larger than the previous works by the order of two. This paper is organized as follows. In the next section, we will illustrate the methodology and data used in this work. The, we show the results of our analysis and discuss the determinants of supplier-customer relationships. And finally, we conclude our paper. Methodology Supplier/Customer finding as a machine learning problem