Hacker News new | ask | show | jobs
Show HN: Convert Text to Query using natural language programming (query.quantleaf.com)
4 points by marcus_pousette 1985 days ago
3 comments

I typed in a dozen queries and never got a result.. Is there a format to follow? What stage of the product is this? The fact I got asked to pay something insulted me since I couldn't find a query that worked.

Maybe, more example queries or "types" of queries would help.

Ok! I understand. Yes, so how it works is that it is your database schema that defines what can be searched on and not. The demos on the website are created from very simple schemas, which means it might not understand what you are looking for in many cases. By reading the suggestions (the words appearing beneath the search field) one can get an idea about what it understands. But it seems that it could be improved!

Thanks for the feedback!

I tried

> A dessert with banana in less than 30 minutes

It detects "dessert" and "30 minutes" but it doesn't detect "banana". Looking at the instructions, I think you are not detecting ingredients, so ... I think this is a feature request.

Is this connected to a real database with many recipes or is only a demo of natural language to a sql-like language?

Yes, this is expected result. It is purely based on your database schema. For the Recipe example, i do not have a field that is called ingredients, that could potentially have the "banana" as a possible value. I can add this to the example.

It is only a demo of natural language to sql-like language. The output viewed is just a preview of the true output, which takes a JSON form.

I want to make interactions between humans and computers easier, by making natural language and computer code indistinguishable. I approached this problem by introducing the concept of ambiguity into a hybrid language with the help of machine learning. This is the first product that uses this technology. Enjoy!