Analytic query answering in a Semantic Data Lake

The management of Data Lake technologies is challenged by the increasing flexibility they provide in data storage, as well as the fast-changing and diverse data they handle. In order to effectively identify relevant sources for analysis, it is crucial to make sense of disparate data, which is especially important in data science applications where users need to analyze statistical measures from multiple heterogeneous sources. In the paper, a knowledge-based approach for a Semantic Data Lake is presented to enable efficient integration of data sources and alignment to a Knowledge Graph, which represents indicators of interest, their mathematical formulas, and dimensions of analysis. A query-driven discovery approach is used to dynamically identify, integrate and rank the sources to respond to a given analytical query.