|
|
|
|
|
by sovreign
2234 days ago
|
|
The "Text to SQL" parser bundled with Allen NLP may be a good starting place (https://demo.allennlp.org/atis-parser). The devil is in the details when you have to make assumptions on what the user wanted though. For instance, does "What was the worst week for tech stocks?" mean by market cap, largest relative losses, etc. If you can constrain the problem you may be able to get away with just named entity recognition. |
|