Mining Collaboration Patterns Between APIs for Mashup Creation in Web of Things

• Main Authors: Mingdong Tang, Yanmin Xia, Bing Tang, Yongmei Zhou, Buqing Cao, Rong Hu
• Format: Article
• Language: English
• Published: IEEE 2019-01-01
• Series: IEEE Access
• Subjects: Web API; collaboration patterns; association rule mining; web of things; mashup; tag expansion
• Online Access: https://ieeexplore.ieee.org/document/8625470/

The Web of Things (WoT) extends the concept of “Internet of Things (IoT)” in that smart devices in the physical world can be interacted with or integrated via popular web technologies (e.g., HTML, HTTP, and Web API). With the WoT, smart devices can use Web APIs to make their data or functionalities accessible by software. With the popularization of Web 2.0 Mashup applications, creating Mashup applications for the IoT (or WoT) via combining different APIs, also has aroused increasing interests. This paper proposes an approach to mining collaboration patterns between APIs to aid mashup creation for the WoT. The goal of the approach is to disclose what kinds of Web APIs are frequently combined together in mashup creation and what kinds of API combination are popular. Based on a real-world mashup and Web API repository, http://PragrammableWeb.com, we exploit the text description and tags of Web APIs and employ an FP-growth-based association mining algorithm to discover popular collaboration patterns between APIs. To overcome the deficiency caused by tag sparsity, the approach also develops a method based on TF/IDF to expand the tags of Web APIs. The experimental results validated the performance of the proposed approach.