The Wikidata Constraints Violations Checker allows you to analyze the number of constraints violations on a list of Wikidata Items. This is useful to better understand which Items need improvements the most and to better understand the data quality of a specific area of Wikidata.
This script requires at least Python 3.6. In your terminal, run:
git clone https://github.com/wmde/wikidata-constraints-violation-checker.git
cd wikidata-constraints-violation-checker
pip3 install -r requirements.txt# To run the script with an input file
python3 checkDataQuality.py -i <inputfile>
# To run the script using randomly generated Item IDs
python3 checkDataQuality.py -r <number of items>
# You can also specify an output filename
python3 checkDataQuality.py -i <inputfile> -o <outputfile>
# Or a batch size
python3 checkDataQuality.py -r <number of items> -b <batch-size>| Arg | Name | Description |
|---|---|---|
| -i | Input file | The path to the file containing the input data |
| -r | Randomly generate Items | The number of Items to randomly generate |
| -o | Output file | The path to the file for output |
| -b | Batch Size | The list of Items are broken down into batches for processing. Default value is 10 |
The script can read CSV files or generate random Item IDs.
Example input file, the first column will be used to query for constrains violations:
Q60,New York
Q64,Berlin
Q70,Bern
Q84,London
Q90,Paris
The following fields are provided in the output data for Items that are succesfully checked.
| Field | Description |
|---|---|
| QID | The unique Item identifier |
| statements | Total amount of statements on the Item |
| violations_mandatory_level | # of violations at a mandatory level |
| violations_normal_level | # of violations at a normal level |
| violations_suggestion_level | # of violations at a suggestion level |
| violated_statements | # of statements with violations |
| total_sitelinks | # of sitelinks on the Item |
| wikipedia_sitelinks | # of sitelinks to Wikipedia |
| ores_score | ORES Item quality score From 1 to 5 (lowest to highest) |
Please be aware that some large Items are skipped during the analysis because the constraint check API times out for them.