Querying web polystores

The database, semantic web, and linked data communities have proposed solutions that federate queries over multiple data sources using a single data model. Nowadays, the data retrieval requirements originating from versatile and broad domains like healthcare and life sciences (HCLS) are changing this conventional trend - of federating query over a single data model - primarily due to the simultaneous use of different data models (CSV, JSON, RDB, RDF, XML, etc.) in a real-life scenario. It's now impractical to assume that the variety (graph, key-value, stream, text, table, tree, etc.) of high volume data residing in specialised storage engines will first be converted to a common data model, stored in a general-purpose data storage engine, and finally be queried over the Web. Nevertheless, in this era where genomics datasets are growing from petascale to exascale, it is now important to exploit such vast domain resources in their native data models. The key approach is to query the vast data resources from their native data models and specialised storage engines. In this paper, we propose a Web-based query federation mechanism - called PolyWeb - that unifies query answering over multiple native data models (CSV, RDB, and RDF). We demonstrate PolyWeb on a cancer genomics use-case where it is often the case that a description of biological and chemical entities (e.g., gene, disease, drug, pathways) span across multiple data models. In order to assess the benefits and limitations of evaluating queries over native data models, we evaluate PolyWeb with state-of-the-art query federation engine in terms of result completeness, source selection, and overall query execution time.