Publication at Faculty of Mathematics and Physics |
2015
Abstract
In this demo we will show aMUSE - a system for automatically translating text, in particular children stories, to simple 2D point and click games. aMUSE consists of a pipeline of state-ofthe- art natural language processing tools to analyse syntax, extract actions and their arguments and resolve pronouns and indirect mentions of entities in the story. Analysed text serves as data the game mechanics operate on, while the story is represented graphically by images the system downloads from the Internet.
The system can also merge multiple stories from a similar domain into a branching narrative. Users will be able to both play games created by aMUSE and create games from their own texts using the aMUSE editor.