Today, Web services are applied in a variety of industries, and compose the building blocks of many Web-based and mobile applications. They are essential for the crossorganizational functional integration and data sharing across the network. On the one hand, how to construct a network in the Web service ecosystem to better organize them is the current research focus. On the other hand, Web service discovery is also a fundamentalforintegratingtherightservicesintothebusinessscenario. Inthiswork,weused a mathematical method for the evaluation of the social web application programming interface (API) network on the basis of data collected from ProgrammableWeb from the perspective of network science. We constructed two Web-API network models. One scale-free network is composed based on Barabasi-Albert model, the other used popularity-similarity optimization model which considers the similarity between nodes to enhance the performance of service discovery. We discussed the theoretical approach thatweusedtoconstructnetworkmodelsandalsopresentthedevelopprocedures. After the two Web-API network models were constructed, we evaluated the two network models, including nodes degree distribution, power-law, exponent, lower bound and preferential attachment. We discovered that Web-API network can suit the power-law distribution, and the performance of service discovery with them are better than typical means.