welcome to arrhythmicity
a virtual reality exhibition space made in dalpofzs

current exhibition: the company it keeps
by: Kevin D. Shabahang

'You shall know a word by the company it keeps' (John Rupert Firth)

we are excited to present Kevin D. Shabahang's work 'the company it keeps'
online until 15.09.2018

you are invited to inhabit a virtual space populated by 2000 words selected from a corpus of english novels. the spatial position and color of the words are expressions of the relationships between them as revealed by the history of their usage (see Kevin's short description below).
now on: the 2000 highest entropy words in our corpus... including words rarely used, unusual spellings, typos and other manifestations from the periphery of our vocabulary.
keep them company


a project made with A-FRAME
(if it doesn't work on your device please try one of the browsers supported by A-FRAME)

Kevin's description of the project:

The present work shows a 4-dimensional semantic space created using the BEAGLE algorithm. The BEAGLE algorithm, first developed by Jones and Mewhort (2007), learns the meaning of words by recording their history of usage within a large corpus of text (see below link for the paper). The present space was constructed using a large corpus of novels.
The original representations were of much higher dimensionality, but we have compressed them to 4 dimensions using the Sci-Kit-learn Multi-Dimensional Scaling (MDS) tool (see link below). The first three dimensions map directly onto the spatial co-ordinates, and the fourth is mapped to colour. Due to computing limitations, we only show about 2000 words, which we have chosen by taking the highest entropy words in our vocabulary. Entropy is an information theoretic measure of the amount of information contained within an event, or in this case, word.

Kevin D. Shabahang, PhD student at The University of Melbourne, Australia.

follow us: facebook twitter vimeo soundcloud