Automatic generation of stories with dialogues

• Main Author: Alcabasa, Lean Alemania
• Format: text
• Language: English
• Published: Animo Repository 2011
• Subjects: Natural language processing (Computer science); Dialogues; Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/etd_masteral/5883; https://animorepository.dlsu.edu.ph/context/etd_masteral/article/12773/viewcontent/CDTG005022_P.pdf
• Institution: De La Salle University

Creative Natural Language Processing (NLP) aims to encompass and model the various aspects of understanding and human creativity in order to design computer systems that can mimic human creativity and provide an engaging interaction with humans. One topic under Creative NLP is Story Generation. As increasing ways of presenting stories are being realized, automatic generation of stories can then find applications in the field of Education and Entertainment. Several existing story generators have already been developed since the 1970s. The work presented here involved the development of an automated story generator that can create stories with both narratives and dialogues. Narratives are used in order to set the setting of the story, as well as to progress the story by defining the events in the story, while dialogues are used to present interactions between the characters, as well as deliver the concepts for vocabulary acquisition in the English Language, which serve as the domain of the stories by having the characters discuss about it. Results showed that the approach of mapping the dialogues to the Actions the characters perform as well as using utterance units to form the dialogues led to the production of stories that received an overall average rating of 2.68 out of 4. Linguists and story writers who manually evaluated 9 stories produced by the system gave an average score of 2.72 out of 4 for the overall quality of the stories, with grammatical aspects receiving an average score of 2.52 while the generated dialogues receiving an average score of 2.80.