3 ms·
This is to build your own knowledge base. In many cases, Wikidata might not have the data you're looking for. For example, in the tutorial I have linked, the ta
by mariarmestre 5y ago
This is to build your own knowledge base. In many cases, Wikidata might not have the data you're looking for. For example, in the tutorial I have linked, the task is to come up with all the products released by a list of companies. Toutiao would be a product of Bytedance. This is a relation that might not exist on Wikidata (I tried to search for it but could not find it https://www.wikidata.org/wiki/Q24835387 https://www.wikidata.org/wiki/Q24835387).
- trystero 5y agoWhy not use that to enrich Wikidata?
- smsm42 5y agoYou can add relationships to Wikidata. Something like "is a product of" probably already has a property, and would be well within the scope of Wikidata.
- yorwba 5y agoI added ByteDance as the creator and owner of Toutiao. (It was already listed as a "product or material produced" on the ByteDance page https://wikidata.org/wiki/Q55606242 https://wikidata.org/wiki/Q55606242 )
- mariarmestre 5y agoWikidata is a very complete knowledge base, but I think there is still room for a tool like this to be used on Wikipedia data. There might be missing information on Wikidata which is still found on Wikipedia (e.g. list of Bytedance products or investors is incomplete, or the number of employees is also missing from Wikidata). This tool can be used to uncover these relationships for your application, or to feed back into Wikidata if it's of public interest. It could also be that you are trying to extract data to train a named entity recognition model. In that case, you want to extract the paragraph or sentence that has the information and the label.