Previous work has explored the structure of the mashup ecosystem, which can be modeled as a network of mashups and APIs. However, it did not offer an explanation for its growth. In this paper, we seek an answer to the question how mashup developers select APIs. One hypothesis that has been put forward is that APIs are selected by their popularity, that is, by preferential attachment. However, this hypothesis is unsatisfying, as it disregards that mashups are composed from multiple APIs, and neglects the learning between mashup developers. Instead, we propose a copying model. We test to what degree developers create mashups by copying other mashups. We show that a good fit between the actual distribution of APIs and our model can be obtained. We can conclude that copying plays a significant role in explaining how mashups are developed. We also identify open research questions raised by the results. Copyright 2010 ACM.

Additional Metadata
Keywords Copying, Evolution, Mashups, Web 2.0, Web APIs
Persistent URL dx.doi.org/10.1145/1944999.1945010
Conference 3rd and 4th International Workshops on Web APIs and Services Mashups, Mashups'09/'10
Citation
Weiss, M, & Sari, S. (Solange). (2010). Evolution of the mashup ecosystem by copying. Presented at the 3rd and 4th International Workshops on Web APIs and Services Mashups, Mashups'09/'10. doi:10.1145/1944999.1945010