Random word retrieval for automatic story generation
Richard Steven Colon, Prabir Patra, Khaled M. Elleithy · 2014
Over the past forty years, significant research has been done on story/narrative generation in which the computer is the author. Many existing systems generate stories by filling in a template or copying an analogous story (and changing the time, place, etc.) or by prompting the user to provide a start to the story. Very few systems generate variable stories without these techniques. While it is impossible to quantify a human writer's inspiration, we can consider a common exercise that authors perform; namely `writing prompts'. A writing prompt is just a topic or idea around which to start writing. The prompt can simply be a few words, which becomes the basis for a story. In this paper we present story generation from the perspective of how human authors create stories via writing prompts. The system will select a few random words as a prompt, which will form the basic parameters for generating a story. But unlike a human writer, a computer cannot intuitively know the context of a chosen word. Therefore, the Internet (and existing `Concept Knowledge' systems) will be used to find the context for the selected words, thus guiding the story generation process.