The world is a multidimensional collage of people, perceptions and places, stitched into the fabric of time. Individual threads intertwining and splicing into senses, perceptions, associations and preferences. How we as individuals experience the world, specifically the (in)finite variety of landscapes constituting our world, not only forms attitudes and opinions, but directly influences our mental and physical well-being. However, we all perceive the world differently depending on our demographics or socio-cultural background. It is thus of utmost importance that we make an effort to understand and analyse subjective perceptions and how these are clustered. Landscape perception and preference research has received unprecedented attention in the past two decades in policy making institutes as well as in the academic sphere. Many scientific groups have contributed valuable insights into how landscapes mould our perceived affordances and consequentially our behaviour. However, traditional methods of data collection are cumbersome, time consuming and expensive. New participatory methods such as crowdsourcing show great potential, however existing datasets are highly biased, shedding a skewed light on how our everyday lived environments are supposedly perceived. In my PhD project I am looking into novel approaches of data generation through an online application called Window Expeditions. Within the application, individuals are encouraged to write, share and explore natural language landscape descriptions. Users are given complete creative freedom in what and how they write, leading to authentic accounts of experiencing the world. After generating a heterogenous multilingual corpus of landscape descriptions, we will use state of the art linguistic and machine learning methods to distil the contributed natural language landscape descriptions into perceptual dimensions (e.g. visual, auditory, olfactory and haptic) to make these comparable.
Let me give you a few examples of landscape perception, as to make it more understandable:
Imagine a rather posh Afro-American woman walking down the streets of New York city. The click clacking of her high heels on the pavement. Her purse swaying ever so slightly, like a pendulum dictating her strides. Now focus on what this individual would hear. Hundreds of voices combining to a blanket of chatter. A distant siren and the constant tooting of car horns. The smell of fresh tar from a construction site. The constant visual overload of billboards and wildly exaggerated screens. These elements of the landscape are perceived by the woman, influencing her behaviour and her comfort. The woman takes out her phone and starts writing a description of how she perceives her current surroundings. She clicks send and the natural language description finds its way to our server on the broadband highway of the modern internet.
Now imagine an old couple visiting the Matterhorn. Arm in arm, walking slowly due to reduced mobility, but enjoying every step. Taking in the scenery with awe, inspiration and being taken aback with an overwhelming feeling of beauty. The couple sits on a bench and marvels at the view. The Matterhorn with its majestic peak reminding us of how small we are. The alpine flowers filling the air with an ever so sweet note. Distant cowbells invoking a feeling of “Swissness” and calling forth mental images of Heidi, Alphorns and fondue. The old couple gratefully take in their surroundings. The woman takes out her smartphone and types in a natural language landscape description. She clicks on send and the message whisks away on the digital highway.
Let’s move on. Picture an Indian man in his twenties working at his spice stand on Khari Baoli in Delhi, standing in front of meticulously prepared upside-down cones of colours and flavours. People walking by, shouting to be heard. The flagrant combination of spices engulfing the street like a mouth-watering fog. The constant rustle of plastic bags being filled with the goods. The powdery touch of the spices. The man takes a short break, whips out his smartphone, types away of how he experiences this landscape and, much like the others, sends his landscape descriptions through the world encompassing web of data to our server in Zurich.
Finally, image a Laotian girl walking beside a rice field, gently holding her mothers hand. She reaches over and glides her hand over the top of the rice plants. She is barefoot and can hear the splashing of her own footsteps. There is distant thunder indicating the daily incoming tropical rain. It is hot and moist, sweat pearls forming happily on people’s skin. A faint smell of a distant barbeque lingers in the air. The mother gives her smartphone to the child who merrily types in what she can hear, smell, see, feel. By pressing send she catapults her experience through space and time to our server in Zurich.
All these mentioned fundamentally different landscapes are perceived and then transformed into an abstract digital representation using natural language. Descriptions from all over the world find their way to our server in Zurich, much like the roots of a global digital tree. In Zurich, these individual roots are combined, compared, dissected, analysed and finally recombined. The individual landscape descriptions (or the roots), are thus, through analyses, reconfigured and combined, into the trunk of the research tree, slowly branching out to individual clusters of similar perceptions.