How to improve the accuracy of intent classification

I am working on a project that uses the GOBOT framework for simple conversations regarding stock informations. E.g., the user will ask a simple question to access the stock price of a company, and the bot gives back the info. Everything is successful except the intent classification part. When I ask the historical price of Apple, gobot will return the answer of realtime price.
image
I am not assigning these intents to slots. The only slot I have for now is company name. And I was expecting that with different dialogues the intent classifier will help me identify what kind of information the user wants.
I check the config file called: gobot_simple_dstc2, and I found that the intent_classifier is null, I am wondering whether that is the reason why intent is not being classified. What value should I use to effectively take use of the intent classifier? I will appreciate very much if you could help me solve this problem!
image

Hey! sorry for the late reply, I started the just started to investigate the case

Sorry for the delayed response, hope you’ll find useful this demo of training an intents classifier!

Thank you for your help, which is very helpful for us to solve this problem !