The enterprise competition in the information society urgently needs the sharing of information resources. However, because there are many different databases within and between enterprises, they together constitute a heterogeneous database system, which makes enterprises face difficulties in achieving the unity, transparency and efficient sharing of information resources. At present, the main method to solve the information sharing of enterprise heterogeneous systems is to use database conversion tools or middleware, but the disadvantages are poor coupling, high cost and complex implementation. The emergence of service-based information sharing mode brings new methods to realize information sharing in heterogeneous systems. This paper makes a systematic and in-depth study of the information sharing architecture, sharing model and key technologies of heterogeneous systems based on services. Decision tree is a method of finding data patterns. It helps to determine the most likely outcome of a given set of events. The working principle of the algorithm is to divide the data into smaller parts, and then analyze one at a time. Once all parts have been analyzed, put them together and form a conclusion about what happened in the past according to their relationship. In this paper, we will use decision tree algorithm to analyze and predict the sharing behavior among employees according to the characteristics of their work environment and the policies of information sharing within the organization.