This paper explores the use of texts that are related to an image collection, also known as collateral texts, for building thesauri in specialist domains to aid in image retrieval....
Abstract. In this paper the suitability of different document representations for automatic document classification is compared, investigating a whole range of representations be...