Textual Data Analysis from Data Lakes

Abstract

Over the last decade, the data lake concept has emerged as an alternative to data warehouses for data storage and analysis. Data lakes adopt a schema-on-read approach to provide a flexible and extendable decision support system. In absence of a fixed schema, data querying and exploration depend on a metadata system. However, existing works on metadata management in data lakes mainly focus on structured and semi-structured data, with little research on unstructured data. Thence, we propose in this thesis a methodological approach to enable textual data analyses from data lakes through an efficient metadata system.