Keywords: Topic analysis, content analysis, foam chart, visual topic exploration.

Topic modeling is one of the most commonly used methods for analyzing the content of text. The hierarchical nature of the output of Dcipher Analytics' topic modeling operation – each topic is associated with a set of keywords with varying degrees of relevance for the topic – makes foam charts a great way of exploring topics and their associated keywords.

Click "Add workbench" and click the Foam Chart symbol. A Foam Chart workbench is now added to the workspace.

If you have already generated topics, drag-and-drop the topic collection from the Schema workbench to the "Display topics" drop zone in the Foam Chart workbench.

If you have not yet generated topics, do so either by running tokenization and topic modeling in global scope, or by dragging the text field that you want to detect topics in to the "Detect topics" drop zone in the Foam Chart workbench. The latter will append operations locally in the Foam Chart workbench and they will not appear in the global pipeline unless they are explicitly moved to global scope.

The topics are now displayed as cells in the foam chart, with associated words as cells inside the cells. Topic cell size reflects aggregated topic strength, while word cell size represents relevance with regards to its topic.

☝️ Note: For the best experience, enlarge the workbench to avoid truncation of topic labels and words.

The Foam Chart workbench offers multiple options for viewing the data, available in the View menu in the workbench header:

To expand a topic and view the words inside it, double-click a topic cell. To return to the topic overview, click the "esc" key on the keyboard.

If, while interpreting the topics, you want to edit their labels, click the "Rename" icon in the workbench header, select your topics label field (typically "topics.labels") as "Text source", select the value that should be renamed under "Value", and type the new name under "New value".