Dezeen Bot List is a data analysis of titles, comments, and images mediated on the web-based cultural platform dezeen.com. One-year’ worth of titles, images, and comments is scraped from the Design category of Dezeen; from April 2016 – April 2017.
The arrows to the left and right of the Bot List site are entry-points to the editorial backend. Words use in Dezeen article titles and words from the comments of readers are listed in chronological order and show frequency of use for verbs, adjectives, nouns, and names. This arrangement suggests the semantic/linguistic character of Dezeen titles and Dezeen commentators.
This backend arrangement shows contrasts between word use in title texts by Dezeen, and the tone of comments left by readers and in response to article content.
The content scraped from Dezeen is recomposed into an image focused structure, titled Dezeen Bot List. The Dezeen Bot re-activates the data scraped from reader comments. Via deep (machine) learning algorithms the Dezeen Bot learns the language and structure of the commentators. The Dezeen Bot responds to images it is shown and generates text responses to these images. The text might be strange, fragmented and ‘dis-informative’ though similar in style, tone and substance to the language of Dezeen commenters; a collective voice.
Images are taged with information through image classification in order for the bot to recognise the image’s content. The classification often reduces and obscures the subject matter of the image and so, an abstracted relationship occurs.