- Tone
- Sentiment
- All Tone categories
- Analytical
- Anger
- Confident
- Disgust
- Fear
- Joy
- Sadness
- Tentative
How to read SunBurst
SunBurst is a radial display of the different topics that are found within the dataset. Topics are categorical labels that are generated from the words within the dataset. The width of each segment of the SunBurst represents the relative number of rows that fall within that topic. The color of the segment correlates with the tone or sentiment category that the dataset is analyzed to posses. The opacity of each segment is determined by the relevance each topic has. Clicking on a segment will render a display with the relative keywords found within those rows.
Use the Comparative Factor selections below the display to redraw SunBurst and see the topic distribution change. The side panel on the right will provide show which individual rows have been associated with each tone or sentiment category.
The SunBurst display technique was conceived in the early 2000s by John Stasko and Eugene Zhang, while together in the School of Interactive Computing at Georgia Tech.
How to read ScatterText
ScatterText distributes keywords (common words found within the dataset) across two different dimensions, Relevance and Frequency. The vertical axis shows the keyword as indicated by its relevance within the category in which it was found. The higher up the vertical axis, the more relevant the word is to its category of Tone or Sentiment. The horizontal axis shows Frequency, how often this keyword is found within the dataset as a whole. The coloring of the keyword’s node corresponds with which tonal or sentiment category it belongs to.
Look for outliers in keywords located in the upper-right quadrant, indicating qualities of both high relevance and high frequency. Isolate a single sentiment category, such as Very Positive or Very Negative, to find out which keywords are commonly being used with negative or positive emotions. Compare these against different factors to see what patterns emerge.
Our interpretation of the ScatterText display is based upon the open-source visualization library by Jason Kessler.
How to read WordTree
WordTree illustrates both the hierarchy and context of individual keywords as they are naturally used in sentences. The larger the word is, the more common it is within your dataset. As you read left-to-right, you’re able to see which words typically follow one another. Navigate in-and-out of branches of the trees to see the details of the sentence as they're revealed.
Vary the applied Social Language, Tone or Sentiment that is being categorized with these sentences to see variances in the contexts of use. Apply factors to see how the hierarchy starts to change when reflecting different usage scenarios.
WordTree was initially constructed in 2007 by Fernanda Viégas and Martin Wattenberg. This technique is also available as a part of the Google Charts library.
Gauge works with Marketing, Product and Research teams to understand experience at scale.