GitTables benchmark - column type detection

<strong>Note: the download page of the entire GitTables corpus is here: https://zenodo.org/record/4943312.</strong> This dataset represents a <strong>small subset</strong> of tables from GitTables curated for benchmarking column type detection methods. This benchmark evaluates systems that match table columns to semantic types from the DBpedia and Schema.org ontologies. It is featured in the SemTab 2021 challenge (CTA task). This dataset consists of the following files: “tables.zip”: directory with a sample of 1101 tables from GitTables. Filenames correspond to table IDs, the first column (without column name) corresponds to row indices, column names are replaced with "col_0", "col_1", etc. which match to the targets and labels (semantic types). “&lt;ontology&gt;_targets.csv”: target columns per table ID, 1 file per ontology (DBpedia or Schema.org): columns are “table_id” (ignore the "_&lt;ontology&gt;" suffix) and “target_column” (i.e. the column that should be annotated). “&lt;ontology&gt;_gt.csv”: ground truth column annotations per table ID, 1 file per ontology: columns are “table_id” (ignore the _&lt;ontology&gt; suffix), “target_column”, “annotation_id”, “annotation_label”. “&lt;ontology&gt;_labels.csv”: unique labels present in the annotated tables, 1 file per ontology: columns are “annotation_id” and “annotation_label”. The labels (semantic types) from each ontology come from: DBpedia: only properties from the DBpedia ontology. Schema.org: properties and types from Schema.org. For the entire GitTables corpus, please refer to this dataset. Visit https://gittables.github.io for more background and contact details.