基于Transformer的服务推荐方法研究

Service Recommendation Method Based on Transformer

摘要: 近年来，越来越多的开发者使用Web服务来进行应用开发，但是如何选择合适的Web对于开发者来说存在着一定的难度。开发者对于各类服务不熟悉，无法精确对服务特征进行描述。因此，本文提出了一种基于Transformer的服务推荐方法(SRT)，首先，我们使用Transformer来对开发者提出的开发需求进行文本特征提取，接着我们使用深度神经网络来进一步挖掘应用和服务的潜在关系，进而进行服务推荐。在ProgrammableWeb上收集的真实数据集上进行的大量实验表明了我们所提出的SRT方法的有效性。

Abstract: In recent years, more and more developers have been using Web services for application development. However, it can be challenging for developers to choose the right Web services. They may not be familiar with various services and find it difficult to accurately describe their features. Therefore, we propose a Service Recommendation method based on Transformer (SRT). Firstly, we employ Transformer to extract textual features from the development requirements that provided by developers. Then, we utilize deep neural networks to further explore the potential relationship between applications and services, enabling service recommendations. Extensive experiments conducted on a real dataset collected from ProgrammableWeb demonstrate the effectiveness of our proposed method.