Mining user behavior pattern in mashup community

Mashup allows users to integrate different kinds of sources together. The ProgrammableWeb.com is a popular online social mashup site that enables users to publish mashups. In this paper, we discover user behavior pattern in mashup community by studying the network and clustering properties of ProgrammableWeb.com. Moreover, we define a new concept-mashup entropy to evaluate the diversity of mashup community. The results prove that mashup community possesses scale-free property. The frequently used APIs and annotated tags attract large amounts of users. People who have similar ideas tend to have similar behavior. Finally, we find that the overall mashup entropy decreases as time goes by, which means that users tend to reuse several design patterns in creating mashups.