The Free Association Database

In my Graduation project for Piet Zwart Institute (Rotterdam, NL) I am studying the interaction between humans and computers. I am specially interested in the current trend - performed mostly by giant ICT companies - to portray technology as invisible as possible.

One of the exercises I did regarding this research was to build a database of freely associated words. These associations were made by visitors of this page, where a word was displayed and the visitor was asked to write the first word that came to his/her mind. The words being displayed were picked up from two different texts: iPad's advertising trailer transcription and iPad's Terms of Service. This device, the iPad, is a concrete example of this pursuit of an invisible technology. It is designed for a user who is not supposed to be curious about the machine. Ideally, the user will forget that the device is a computer and will regard it as 'a magical pane of glass'.

One of the possible outcomes for this database will be to serve as content for an interactive installation where user and computer communicate through a voice interface.

The graphic below shows the map of built relationships. It can be explored either by scrolling the page and hovering the mouse over the words or by clicking on the START button. In this case, a series of relationships will be displayed on screen, revealing also the sequence of the words. The first word is always one that came directly from the texts mentioned above. The last word in the sequence is always a visitor input. The words between the first and the last were both visitor input and retrieved from the texts. The sequence of the words in the relationship cannot be seen by hovering the words. On the other hand, all connections to/from that word are highlighted.

Each visitor's contribution was saved in a text file containing the pairs of words. This content was processed using Python to generate a file in JSON format containing the words (nodes), the relationships (links) and the weight of each word (the number of times this word was related to another). This JSON file was then used to display this information graphically, using D3.js

As of March 14 2015, the visitors produced 156 text files, in which 3452 connections were made among 2675 words. When a connection between two words was made more than once, that is indicated through the opacity of the line, which is higher. Also, the number of times which a word has been used (its weight) is reflected on the size of the circle representing that word. The words were processed 'as is' - no cleaning was performed, except for replacing an empty field with the word [empty].