Training over context and obtaining answers about it

[AnibalDuarte]

Is your feature request related to a problem? Please describe.
My problem is my lack of understanding of the NLP itself.

Describe the solution you'd like
A tip to how to train the chatterbot

Describe alternatives you've considered
I'm new to this life of training robots and I wanted to know if NLP has a solution for my specific situation:

• In a certain online game players get a certain currency by killing monsters. I want to send to the BOT all occurrences of the "drop" of this currency containing: the character's class, its level, its gearscore, the amount obtained and the complete date.
• In that same game the characters can put items in the auction. I want to tell NLP what the minimum initial sales values are for items, and then progressively tell them when items go on sale, what their gearscore is, when they sell and how much they sold for, and who bought them or when they were cancelled.

The idea is, after accumulating the data, to be able to ask which classes manage to obtain more coins, what is the average obtainment per level and per gearscore per hour, as well as asking which items have the highest turnover in the auction and, also, if there are players who recurrently buy and sell with each other.

Extra context
After reading more about it I've realized I need to input the answers on the model too, now my question is how I teach the data itself then request the answer as data as well (for instance, as JSON)?