The explosion of reusable Web services (e.g., open APIs, open data sources, and cloud/IoT services), has become a new opportunity for modern service-composition based applications development. However, this enormous growth of Web services increases the difficulty of selecting the best suitable Web services for a particular application. Hence, the design of an effective and efficient Web service recommendation, primarily based on user feedback, has become a challenge. In the mashup-API recommendation scenario, the most available feedback is the implicit invocation data, i.e., the binary data indicating whether or not a mashup has invoked an API. Various efforts are exploiting potential impact factors, such as the invocation context, to augment the implicit invocation data with the aim to improve service recommendation performance. One significant factor affecting the context of Web service invocations is geographical location, but it has been given less attention in the implicit-based service recommendation. In this paper, we propose a probabilistic matrix factorization based recommendation approach, which considers geographic location information in the derivation of the preference degree underlying a mashup-API interaction. The geographic information, which is integrated with functional descriptions, complements the mashup-API invocation data input for our matrix factorization model. We demonstrate the effectiveness of our approach by conducting extensive experiments on a real dataset crawled from ProgrammableWeb. The evaluation results show that augmenting the implicit data with geographical location information increases the precision of API recommendation for mashup services.

可重用Web服务（例如，开放的API，开放的数据源和云/ IoT服务）的爆炸式增长已成为基于现代服务组合的应用程序开发的新机会。但是，Web服务的这种巨大增长增加了为特定应用程序选择最合适的Web服务的难度。因此，主要基于用户反馈来设计有效的Web服务推荐已成为一项挑战。在mashup-API推荐方案中，最可用的反馈是隐式调用数据，即指示mashup是否已调用API的二进制数据。各种努力正在利用诸如调用上下文之类的潜在影响因素来增加隐式调用数据，以提高服务推荐性能。影响Web服务调用的上下文的一个重要因素是地理位置，但是在基于隐式的服务推荐中却很少关注它。在本文中，我们提出了一种基于概率矩阵分解的推荐方法，该方法在推导mashup-API交互所基于的偏好度时考虑地理位置信息。地理信息与功能描述集成在一起，对我们的矩阵分解模型的mashup-API调用数据输入进行了补充。我们通过对从ProgrammableWeb爬取的真实数据集进行广泛的实验来证明我们的方法的有效性。