Ask questions from Data

Convert NLP to SQL

1. Check NLP statement. Convert these into tokens. From statement derive intent of NLP statements.

2. Based on intent determine which table user want to get access, what columns user is interested in, what condition on data should be applied e.g. date ranges.

3. Determine whether aggreegate functons e.g. sum, avg, corr need to be applied

4. Determine where model has high confidence vs low confidence

5. Generate SQL

Above is flow for converting one NLP statement to SQL statement.

Passing context across statement is even harder.