Content improvement implementation

Following is a description of the transformation processes for SRC-AP assets that are edited in VocBench and shall be published as SKOS-AP-EU in Cellar. The current implementation of the transformation pipelines uses Linked Pipes ETL (https://etl.linkedpipes.com/). The transformation processes aims to clean up the Authority Tables (AT) content using a well-established framework to resolve semantic dissonances, redundancies, overlaps, and other types of issues.

Folders:

• SAI_W.P2.1_pipelines contains the pipelines with all executed transformations that aim to simplify the Authority Tables.
• Old_Pipeline contains all the developed process for previous publication process of Eurovoc.

Input:

• SRC-AP files for Corporate Body, Corporate Body Classification, Country, Membership Classification, Site, Language and Place Authority Tables.

Output goals:

• To generate simplified and clean SREC-AP files.

ETL pipelines description

We developed a structured approach to implement the recommended actions based on the findings and executed transformations on the ATs in their SRC-AP representation, as following.

• Extract the data from a GraphDB RDF database (triplestore), where the SRC-AP files are uploaded.
• Apply transformations within the pipeline to convert the source data into the desired target representation.
• load the processed data back into the triplestore in a new target environment, ensuring that the refined data would be readily available for use.