Michael Young
Michael Young
Yes the interface is a bit slow. This is a prototype not a production ready model. There are multiple calls to wikidata/dbpedia.
Currently, entities are resolved by taking the first result returned from WikiData's entity search so China will resolve to the country. I may work on a feature that will return...
NLQuery uses python2.7 and I haven't had time to add backwards compatibility from python3. There are tools for upgrading NLQuery or you can switch to python2.7.
This is currently Wikidata specific because I only wrote adapter for WikiData. WIkiData has a really good entity search which resolves strings to entity IDs. I found DBPedia's equivalent entity...
Currently, this system is English only since the parse tree are in English and the rule matching is only in English.
Yes it can be made dynamic, you just need to change the inputs/outputs
Yes I've applied this kinda of approach to answer questions with data sourced from WikiData and DBpedia. Look at my other repo: [NLQuery](http://github.com/ayoungprogrammer/nlquery). There are more sophisticated information retrieval methods...
https://github.com/ayoungprogrammer/grafana-finance/blob/master/src/datasource.js#L8 it should be `quandl_api_key`
The latest version of the Quandl API has /datasets as the time-series data endpoint and /datatables for the tabular data endpoint. I believe tabular does everything time-series does but provides...
Ok so I successfully added a special case for the WIKI Quandl code, but unfortunately the datatables API requires CORS. You can retrieve the WIKI data but would have to...