My first requirement was creating an assistant which can be fed documents and the can be trained to answer questions related to the contents inside the documents, something very similar to this issue Document Classification. But it seems that it is not possible(correct me if I am wrong) to convert normal pdf/text based document into trainable yml files.
So now, in order to simplify the needs, I am thinking to design an assistant such that it can fetch a question and in the background the data base contains manuals/files related to a project and each project contains multiple version of the manuals/files. So, the assistant should ask what project and then what version and then based on that it should find the context and return the document with an abstract of the context. Is this achievable using rasa? along with the use of some other searching mechanism?. I am new to rasa or even to the development of assistants, any help will be greatly appreciated.
Thank you,
Yash Chopda