Accessing Web APIs as "Virtual" Relational Databases or Knowledge Graphs

Query languages for databases (e.g., SQL) and knowledge graphs (e.g., SPARQL) provide developers, analysts and researchers with a concise, declarative, and highly flexible mechanism to access (or even modify) the data of interest in a large dataset. Yet, in many use cases part or all of the data is accessible only through Web APIs, which makes data access comparatively less convenient, less flexible (due to restrictions on available API methods), and possibly slower enough to be incompatible with users interactively querying data. In this session, we present open source components that tackle these issues by enabling querying a Web API as a “virtual” (since not materialized) relational database via SQL, or as a “virtual” knowledge graph via SPARQL, at the same time providing pre-computation and caching solutions of Web APIs results to speed up data access. The core components presented in the session have been developed in HIVE, a local “Fusion Grant” research project involving UNIBZ and Ontopic Srl that aims at extending virtual knowledge graphs to Web APIs, with special focus on NLP APIs (for accessing unstructured data) and on applications in the domain of environmental sustainability.