Sending the data from one model into another is called "chaining." Chaining allows you to get even more detailed output than you could with just one model.
To chain models together, you first need to think about what each model is good at. For example, if you have a model that can identify objects in an image, and another model that can identify faces in an image, you can chain them together to get even more detailed output.
To do this, you would send the data from the first model (the one that identifies objects) into the second model (the one that identifies faces). This way, the second model can use the data from the first model to get more accurate results.
Your first model is a language model that detects if a person is being mean or nice.
Your second models are used based on mean or nice, if mean is greater in value than nice ( say .7 nice, .3 mean ) then the data gets sent to the mean model
the mean model is used to determine what kind of mean statment has been made: evil, funny, malicous, etc..
if nice is greater in value than mean ( say .2 mean, .8 nice ) then the data gets sent to the nice model
the nice model is used to determine what kind of nice statment has been made: sweet, supportive, etc..
at the end of this you have 2 different models that give you more information about the data you put in.
you can now use this determination to send the data on to yet another model to determine the intent of the statement or to do something with the data.
model chaining is a great way to get more information from your data and to make your models more accurate.