Within the framework of the research fellowship at the University of Antwerp, I presented my research so far to a conference on Digital Humanities in Utrecht.


Paper Extract

Stories play a vital role in the lives of children. The alternative worlds they produce encourage imagination and creativity, but also transform knowledge into structures that children can understand and relate to. We present an interactive story system that creates narratives from Wikipedia articles, and reveals them through dialogue with a user. Using state-of-the-art narrative generation tools and a chatbot dialogue system, information from Wikipedia is revealed to the child based on their input. Generating narratives from any text has long been a goal of Artificial Intelligence researchers because narrative structures are useful for learners of all ages. However, many of the existing story generation systems have relied on hand - written techniques that cannot meet the scale of data online. In recent years search - based systems have been able to incorporate broader topics, but they have sacrificed continuity, producing fragmented narrative. Here we propose a search - based system that scours Wikipedia for articles relating to user input, and then restricts its generation material to that article; in doing so, the system utilises the defined topic and chronology of the article to retain context.

Full example