Hello all,
I have a dataset with two columns one of them is a prefix and the other is a suggestion. Eg: Prefix :University of M, Suggestion: University of Montreal. I trained a seq2seq model to achieve this task. But seq2seq models are more general and there can be many number of suggestions for the prefix 'University of '. I was able to extract a set of words from the dataset that provide context. I want to use this set of words to provide the model with context and some how I want the model to prioritize choosing from this set. The context words are all of different nature (Norcross, Trump, Football etc). I don’t have any labelled data to predict slots. Is there any way I can achieve this task?
Thanks in advance.