Logo des Repositoriums

Auflistung nach Schlagwort "ETL"

1 - 2 von 2
  • Konferenzbeitrag
    Converting data organised for visual perception into machine-readable formats
    (44. GIL - Jahrestagung, Biodiversität fördern durch digitale Landwirtschaft, 2024) Aue, Alexander; Ackermann, Andrea; Röder, Norbert
    Spreadsheets are used to store an extraordinary amount of important data. The fact that spreadsheets are both easy to use and allow users a great deal of flexibility in how they store their data is a significant reason why they are so popular. Users often use a variety of layout techniques to make the data easy for humans to understand. But this layout also creates problems for traditional Extract-Transform-Load (ETL) tools. We propose a program that allows users to easily extract data from Excel files by selecting the cells containing the data and metadata thereby determining the data hierarchy. We have used this program to extract data of the Agricultural Structure Survey on land use and livestock in Germany, which does not follow a nationwide standard, leading to large differences in the structuring of the data between the federal states, making it a good benchmark.
  • Konferenzbeitrag
    Harmonizing OER metadata in ETL processes with SkoHub in the project "WirLernenOnline"
    (Proceedings of DELFI Workshops 2022, 2022) Rörtgen, Steffen
    The values for metadata attributes of Open Educational Resources (OER) are often made available in repositories without recourse to uniform value lists and corresponding standards. This circumstance complicates data harmonization when OERs from different sources are to be aggregated in one search environment. With the help of the RDF standard “SKOS” and the tool “SkoHub-Vocabs”, the project "WirLernenOnline" has found an innovative, reusable and standards-based solution to this challenge. This involves the creation of SKOS vocabularies that are used during the ETL process to harmonize differing terms (for example, "math" and "mathematics"). This then forms the basis for providing users with consistent filtering options and a good search experience. The created and open licensed vocabularies can then easily be reused and linked to overcome this challenge in the future.