Return to search

Procedural Narrative Generation Through Emotionally Interesting Non-Player Characters

Procedural content generation is a technique used to produce a wide range of computer-generated content in many industries today, the video game industry in particular. This study focuses on how procedural content generation can be applied to create emotionally interesting non-player characters and through this, generate narrative snippets that can immerse and interest a reader. The main points examined are how to achieve this using a modular approach to personality and behaviour, how well readers can distinguish whether motivations and interactions are generated by a computer or written by a human, and to what degree a reader can be immersed in a computer-generated narrative. Procedural narrative could help to reduce workload on large projects or lower costs, and is an area in which there is much room for further research. To answer these problems, a literature review of existing techniques for the creation of emotionally interesting non-player characters was conducted and used to design and construct a prototype implementation for generating procedural narrative. The output of this narrative was dressed up to match the style of a human text and A/B testing was conducted utilising a survey in order to evaluate and compare responses to the two texts. Ultimately, the results showed very little difference between the perception of the human-written text and the computer-written text, with the only aspects found lacking in the computer-written text being clarity of emotion and foreshadowing.

Identiferoai:union.ndltd.org:UPSALLA1/oai:DiVA.org:lnu-76708
Date January 2018
CreatorsGriffith, Ioseff
PublisherLinnéuniversitetet, Institutionen för datavetenskap och medieteknik (DM)
Source SetsDiVA Archive at Upsalla University
LanguageEnglish
Detected LanguageEnglish
TypeStudent thesis, info:eu-repo/semantics/bachelorThesis, text
Formatapplication/pdf
Rightsinfo:eu-repo/semantics/openAccess

Page generated in 0.0016 seconds