A Comparison of Unsupervised Methods to Associate Colors with Words G¨ ozde ¨ Ozbal † , Carlo Strapparava † , Rada Mihalcea ‡ , Daniele Pighin § † FBK, Trento - Italy ‡ UNT, Denton TX - USA § UPC, Barcelona - Spain Abstract. Colors have a very important role on our perception of the world. We often associate colors with various concepts at different levels of consciousnes and these associations can be relevant to many fields such as education and advertisement. However, to the best of our knowledge, there are no systematic approaches to aid the automatic development of resources encoding this kind of knowledge. In this paper, we propose three computational methods based on image analysis, language models, and latent semantic analysis to automatically associate colors to words. We compare these methods against a gold standard obtained via crowdsourc- ing. The results show that each method is effective in capturing different aspects of word-color associations. 1 Introduction Colors have a significant effect in our daily lives and the way we perceive the world is strongly connected to their presence. Several psycholinguistic studies [9, 1, 7, 15] have demonstrated a deep connection between colors and emotions. This connection is generally not straightforward, in the sense that it is not easy to directly map colors to emotions. On the other hand, we typically build associations between colors and words and, in turn, between words and emotions. As an example, by establishing a connection between black and darkness or death, and by observing that these two words are often associated with fear, we could automatically link the color black to fear. In this respect, the investigation of the relation between words and colors is a very important step towards the automatic connotation of the emotional content of a word. However, to our knowledge, there is no resource that contains information about the association of words and colors. Such a resource could be important for many applications. For instance, it could be useful for several educational purposes such as attracting attention, facilitating the memorization process by triggering the visual memory in addition to the verbal memory [14]. This resource could also be used for advertising by automatically coloring advertisement texts based on the appropriate colors, according to the common sense knowledge or the emotion that we wish to convey. Word-color associations could also be used as features in algorithms for automatic image tagging [10]. In this paper, we propose three methods to automatically associate colors and words in English, and compare their performances. Each of these three methods