In the Margins

The Multicloud Tool

Word clouds are a method of visualizing the Document-Term Matrix. They present terms arranged at angles for compactness, with each term sized according to its frequency within the text. Word clouds enable you to get a sense of the content in your corpus, and they are very good for presentations. However, they also have some well-known limitations (see the topics article on visualizing texts with word clouds). In some languages, individual tokens may not correspond to words, which will limit the usefulness of this method of visualization.

The Lexos Multicloud tool uses allows you to use word clouds to generate a comparative visualization of your documents. Each document has a separate word cloud, displayed in a grid that allows you to “eyeball” similarities and differences between the proportions of terms in each document. Because the word clouds are necessarily small, they may not always be easy to interpret. If you find interesting documents using the Multicloud tool, it is recommended that you also generate larger clouds of each individual document using the Word Cloud tool.

Word clouds are generated from lists (vectors) of term and count values, where the proportion of the overall counts determines the size of the terms in the layout. This method of visualizing term vectors is useful as a basic strategy for interpreting the results of topic models, a common method of performing computational textual analysis not offered in Lexos. In the Humanities, the MALLET topic modeling tool is commonly used, but it does not produce output that is easily visualized. The Lexos Multicloud tool allows you to upload MALLET data and convert each topic into a Lexos document, from which you can generate word clouds or perform other operations such as cluster analysis. Note: Generating topic clouds from MALLET data has been removed in v3.2 of Lexos. We will restore the feature in v3.2.1. If you have need of this feature, you can download v3.1.1 and install it on your local computer.

Generating and Using Multiclouds

Select some or all of them using the Select Document(s) check box. Then click the Get Graphs button. Please keep in mind that generating multiclouds is computationally intensive and may take a very long time if you are generating a large number of clouds. Your browser may hang for some time. Please be patient. Once the graphs are generated, you can run your mouse cursor over each term in the word cloud to generate a tooltip showing the number of times the term occurs in the document.

Rearranging Multiclouds

If you generate more than 6 word clouds, you may find that you cannot view them all on the screen at once. In order to facilitate the comparatison of multiple word clouds, the Multicloud tool allows you to drag and drop individual clouds into new positions in the grid. For instance, you can drag cloud 20 to the top of the grid in order to view it next to cloud 1.

Exporting Multiclouds

The Multicloud tool does not yet have an export function, so the best way to save the Multicloud display is with a screen shot. Some screen capture programs such as Fireshot allow you to capture the entire page, rather than just the visible area, which will allow you to save an image of all the multiclouds. In Mac OSX, you can click Command-Option-F for full screen followed by Command-Option-3 or Command-Option-4 .

This page has paths:

Contents of this path: