A study on methodologies of story generation / Muhammad Riza Abd. Aziz

Storytelling has been a very important human activity since the development of language. We tell and receive stories everyday in the form of poems, novels, movies, songs and plays. Story generation attempts to mimic computationally-achievable aspects of human storytelling in order to create programs that can tell stories. The benefits are numerous: more immersive games, interactive fiction, computer-assisted art and even contributing towards the development of a true artificial intelligence. This study aims to combine findings from literary theory with story generation algorithms in order to investigate computer-based storytelling. After examining the methods of story generation employed by three major programs (Talespin, Gameworld and Erasmatron) a hybrid character-driven method was chosen for implementation in a model program called Crooks, The program also includes some insights gleaned from literature theory (namely Georges Polti's theories) to help create more complete stories.