The Shape of Data
Data can be gathered in many ways, but how is it accessed? And what does it reveal?
As part of our Sense of Here project we wanted to find out more about people's sense of connection to their local green spaces, to the Lake District National Park, and to the natural world in general. The Data of the Heart Survey received over 230 responses and revealed strong feelings.
The survey, using an interactive mapping platform created for us by ESRI, drew in some very thoughtful responses. One of our challenges was to find a way to share the depth of feeling revealed through individual responses and the trends: what matters most to people?
We turned to Vatsala Nundloll and Stephen Mander of the Ensemble Programme at Lancaster University. As software experts they were able to come at the challenge in a way we could never do. Vatsala and Stephen drew on existing programmes and created bespoke coding to enable analysis of the responses: hundreds of responses, thousands of sentences, and a huge range of sentiments.
How does it feel?
The process has resulted in a visualisation of data that's interactive and multi-layered. Vatsala Nundloll used Python and Jupyter notebook to write a code; and Scattertext to generate these visualisations. The graphs below are a selection what we have created from the survey results: they offer a way to access the complex written answers and summarise important issues and feelings.
To go in further, click on the image to open the html file - this allows you to follow your curiosity. Click on an individual word, or a coloured dot, and scroll down to read the relevant responses which will appear below the graph. Or use the search bar to type in a word that you're curious about (e.g. farming, wildlife, transport ... anything really) and discover where and how it has been used by participants.
How can we measure hope?
Stephen Mander was particularly interested in finding a way to anlyse the sentiment behind the responses - to show the balance of negative or positive emotions, in the context of each survey question. Using Python and linking to a vast library that detects individual words and interprets language fragments and sentence structures, Stephen plotted the quality of emotion, on a positive-negative scale. It's still early days with this work and we are continuing to develop and refine it, but with these early results it's good to see, in the context of the level of hope for the future of the environment, that there's a tendency towards optimism.
Can a computer write poetry?
Stephen has also set himself the challenge of writing code so that a computer can generate poetry from the responses. There are coding systems currently in development that allow AI (artificial intelligence) to write articles that may seem to be written by humans. Can a computer, though, write poetry? Stephen has devised coding that helps a programme select words and word pairings, and bring them together. The results are pretty random. We like his thoughts on computer generated poetry:
"Poems are far more than rules: they're art. Perhaps it's cruel to call these works poetry as much as just words for inspiration because they don't contain knowledge of what they speak of. I've grown to appreciate that data, language and the way humans want the two to interact are messy concepts. Somewhere in the mess, there are some beautiful ideas."
Harriet gave the computer-generated 'poems' a bit of editorial polishing ... here are three short pieces arising from the computer's interpretation of survey responses.
Childhood is a wide local, a wild, open identity
listening like land, searching for the inaccessible season
for silence, freedom and peace.
Never dislike a walk - you will see:
the mountain provides the path
to tranquillity and awe.
Life is a green space, but loss?
Loss is a quiet management,
Beauty, a quiet vandalism,
A lonely silence.
Huge thanks to the Ensemble team. You can find out more about the work of this 5-year Fellowship programme that brings many disciplines together here: https://www.ensembleprojects.org.