By Marian Dörk and Dawn Knight carried out when at Newcastle University
funded by the RCUK Digital Economy programme’s PATINA project.
More details about the concept and methods are in this article.
The code of the WordWanderer prototype is on GitHub.
The WordWanderer interface has three main views:
The cloud view displays the most common words of a text using font size to represent relative frequencies. The ordering of the words is alphabetical from top to bottom, left to right. When hovering over a word, its relationship with other words, in terms of their rate of proximal co-occurrence in a text, are highlighted in colour.
Clicking on a word selects it for the context view. The word sizing corresponds to how often they occurr in the text near the selected word: the larger the word, the stronger the association between the words and the selected term. Vertical positioning is alphabetical and horizontal positioning represents sentence positions relative to the selected word.
Dragging a line between two words triggers the comparison view. In this view a word's size corresponds to the combined association strength with the two words. Here, the horizontal position represents the difference in association between the two selected words. If a word occurs more often with the left word, it will be placed more towards the left side of the screen.
All views show a limited a number of words. To find a specific word there is a search function above the visualization. Simply type in a word and hit enter to activate this function. In the options (top right), you can set which types of words are included, based on their parts of speech. When in context view, the instances of the selected word are displayed below the visualization.
Show (select all)