Mashup-oriented API recommendation via pre-trained heterogeneous information networks

Abstract

Combining different Web APIs to create Mashups has become very popular nowadays. Choosing suitable ones from massive Web APIs is of vital importance for efficient Mashup creations. A number of Mashup-oriented API recommendation methods have been proposed to address this issue, but they have limitations in their ability to exploit the rich attributes and connection data of Web APIs, which impedes their performance. By modeling the API-related data as a heterogeneous information network and using pre-training technology, this paper proposes an accurate API recommendation method, named PHRec. In this method, the meta paths of APIs in the heterogeneous information network are exploited to obtain their context semantics; the method adopts an attention mechanism. Extensive experiments have been conducted with a real Web API dataset to evaluate the proposed method. The experimental results demonstrate that it significantly outperforms the state-of-the-art methods in the Web API recommendation task.