A-MaGe: Atomic Mashup Generator for the Web of Things

Abstract

Individually, Internet of Things (IoT) devices are often not able to achieve complex functionalities and, therefore, need to be composed together into useful mashups. However, given the current fragmentation of the IoT domain, designing a mashup is still a manual task that is time-consuming and error-prone. The introduction of the Thing Description (TD) from the World Wide Web Consortium (W3C) is meant to facilitate the interoperability between IoT devices and platforms by providing a standardized format to describe the network interfacing of entities, called Things, participating in the Web of Things (WoT). Furthermore, the System Description (SD) extension introduces the notion of Atomic Mashups (AMs), small mashup building blocks that are easier to design. However, designing AMs remains a manual task and, given the rising complexity of IoT devices, manually exploring the resulting design space is infeasible. In this paper, we introduce A-MaGe: a method and its open-source implementation that takes the TDs as an input and uses predefined templates, user-configurable rules, semantic annotation filtering and natural language processing to automatically explore and reduce the design space. SD-compliant UML Sequence Diagrams of the resulting mashups are presented to the human agent for further selection to generate the SD of the mashup as well as implementation code based on the W3C WoT Scripting API. We show that the generation process is fast, allowing multiple iterations by the human agent to increase reduction and we evaluate the filtering power of different filters and constraints. Thus, in combination with the TD standard, our method ensures easy composition of services in heterogeneous environments.