Pattern Mining to improve existing model

Does anyone have recommendations for how to go about pattern mining a database and feeding the recognized patterns to my working openai model (its current functionality is generating queries out of natural language that are then executed on the db). I am looking to make my model more robust so that it can be used not only for answering numerical questions about records in my db, but also use mined patterns to give more insight on the data.


Do you have any ideas on the kinds of patterns you want to find? The issue as I see it, is that your database contains more information than the model can view in one go, you might be able to split the data into chunks and then have the model parse the chunks to look for patterns, but having some idea of what you want to look for is helpful.

1 Like