Polly: A Language-Based Approach for Custom Change Detection of Web Service Data

An ever-growing number of web service providers expose data that is continuously changing. Use cases arise where being notified about changes made to the data is essential to the client, for instance to know when a user has a new follower on Twitter. Monitoring changes on web services data consists in polling services for the required data, detecting any changes in the targeted data subset, and notifying the user only about the relevant changes. However, each step of this process can be relatively complex, leading to a tedious and challenging implementation for developers. In this paper we introduce Polly, a domain-specific language for describing change detection strategies in JSON data fetched from REST web APIs. By leveraging the domain knowledge of the user, our domain-specific language offers declarative, concise yet highly-expressive constructs for specifying change detection strategies. We validate our approach using several user-driven scenarios provided by our industrial partner and show that it outperforms the state-of-the-art solutions.