Hi everyone,
Is Wikidata high quality? What items are due for promotion to the showcase? Answering these questions would require reviewing millions of items on a regular basis and that would take far too long.
So we're training machine learning models to evaluate the quality of items in Wikidata automatically. In order to train these machine learning models, we need to label a bunch of items by their quality level [1]. I'm writing to invite you to join our working group.
We're trying to get ~5,000 items labeled. We'll use this labeled data to train a machine learning model for ORES [2], and use it to categorize the rest of items in Wikidata, look for trends & coverage biases, etc.
If you want to help us train the model, you can navigate to our labeling campaign information page [3] and sign up list [4].
Many Thanks,
Glorian, Aaron, Amir