This study, done in 2013 by Emanuela Haller and Traian Rebedea, tries to design a chatbot that simulates a historical figure, much like I’m attempting with the Historic Voicebot. They aimed to design a chatbot character with a matching personality, by only using data gathered about this historic person from websites and plain text sources.
To extract facts from a text, a set of different digital tools were used, including the Stanford CoreNLP libraries and WordNet. Not only did they use facts about their chosen historical figure, basic information about all the other people that were mentioned was also added to paint a complete picture.
It was interesting to read which resources were used and how they were able to turn a plain text into facts and those facts into a conversational chatbot. Although the technologies used might be outdated, the steps outlined in their research don’t seem to have changed much, making this a valuable resource.