The Sensorial Map of the City

Our daily urban experiences are the product of our perceptions and senses [7], yet the complete sensorial range is strikingly absent from urban studies. Sight has been historically privileged over the other senses and urban studies often focused on the visual dimension of cities [4, 3]. However, smell and sound have a huge influence over how we perceive places, they impact our behavior, attitudes and health. Yet, city planning is concerned only with a few bad smells and with limiting noise levels. Not knowing what smells and sounds exist in cities may result in partial views of the collective image of our urban areas, the proliferation of clone towns (as smell and sound contribute to the city’s identity [1, 9]), and in reinforcing socio-economic boundaries, as smell and sound that provide insights into the social life of cities are used (often unconsciously) as invisible markers between neighborhoods with different socio-economic status [5]. In our work we propose a new way of capturing nuanced olfactory and auditive perceptions of cities from data implicitly generated by social media users. The idea is to search for smell and sound-related words on social media content. To this end, we first build two dictionaries of smell and sound-related words. We ventured out in the urban world and conducted “smellwalks” in a variety of cities: participants were exposed to a range of different smellscapes and asked to record their experiences. As a result, smell-related words have been collected, classified, and enriched with words available from previous studies on smell [2], thus creating the first dictionary for urban smell. Similarly, to obtain a sound-related dictionary, we collected soundrelated words from previous literature on sound and from crowdsourced online data. For both Barcelona and London, we collect geo-referenced picture tags from Flickr and Instagram, and geo-referenced tweets from Twitter. We match those tags and tweets with the words in the smell and sound dictionaries. To create a structure for a large and apparently unrelated dataset of words, we cluster the co-occurrence