Peter Gorniak and Deb Roy. (2004). Grounded Semantic Composition for Visual Scenes. Journal of Artificial Intelligence Research, Volume 21, pages 429-470.