22 Dezember 2016 Engineering

Tag that Cloud: A New Visualization in Kibana

Von Thomas Neirynck

The 5.1 Elastic Stack arrived last month, and with it, a brand new and svelte visualization for Kibana: Tag Cloud! It is the first new visualization in a long time and we look forward to our users using Tag Cloud to enrich their dashboards, reports and presentations.

Introduction

Tag or word clouds show a collection of words, terms or small phrases, laid out all adjacent to each other. The size of the tags or words generally corresponds to their importance. In Kibana specifically, importance here means whatever metric you chose to calculate for a term aggregation. Often this would be a count of most prominent terms, but any other metric supported by Elasticsearch can be used to size the tags.

Word Cloud - Countries

Countries with the most visitors to the elastic.co website, using a sum aggregation, based on a sample of recent traffic.


You might have seen tag clouds accompany newspaper articles about the occurrence of prominent or unusual words in political speeches, or in online databases for collections of relevant keywords. Or here below, where two clouds provide a quick overview of the most common home worlds and weapons in the Star Wars universe.

Word Cloud - Star Wars - Weapons
Word Cloud - Star Wars - Homeworlds
Tag Clouds are then suitable infographics for when you want to convey the most important themes and threads of your "story," and do so in highly visual and legible manner.


Creating your own tag cloud

The options in the design panel will be familiar to frequent Kibana users.

Users can select the easing function that maps metric value to font size. The default is a linear 1-to-1 mapping. Alternatively, use the log-scaling if you want to downplay differences in metric values. In practice, this one means that the difference in font-size will be smaller for higher values. Finally, use the square root easing function if you want to exaggerate differences in metric value. In practice, that one means that the difference in font-size will be smaller for lower values.

For those who like their buffalo wings mildly flavored, the default will generate clouds where all the words are upright. The daring can spice up their clouds with more orientations. Know your audience of course, and choose wisely.

Word Cloud - Shakespeare Characters 1
Word Cloud - Shakespeare Characters 2
Word Cloud - Shakespeare Characters 3
The most blithering characters in Shakespeare's plays, using three different layouts.


In some situations, there will not be sufficient space to position all the words on screen. This occurs when the tags have too large a font, the words are too long, or there are just too many tags to begin with. Kibana will omit tags to ensure there are no overlaps.

To avoid these omissions, provision sufficient space for the Tag Cloud to show all the words.

Word Cloud in a Kibana Dashboard - 1

Kibana omits tags on the left because the container is not large enough to contain all ten of the tags. The same cloud on the right does have sufficient space and displays all the tags.


Another alternative is to limit the size of your clouds to a lower number of terms, using the size parameter in the data panel.

Adding to a dashboard

Just like all other Kibana Visualizations, you can add a Tag Cloud to a dashboard. And just like on any other dashboard, you can interact with Tag Cloud to add filters to the filter bar.

Windows 8 Web Traffic
Selecting only web traffic from visitors using Windows 8.


Pie charts or a simple tables are often used for that same purpose. Now, Tag Cloud serves as a more playful alternative for your end users to drill down in your data.

Naturally, Tag Cloud integrates with the rest of the Kibana family. Install X-Pack to include tag clouds in your reports, save and restore them in the Object Manager, and use them in Kibana's Elastic Cloud deployments.

Feedback please!

With Tag Cloud, after some hiatus, we hitched another wagon to Kibana's Visualization train. But do not presume this is the last one though, or that it will take another year for us to introduce new visualizations. We have big plans for Kibana Visualize and are working on some major new additions and improvements in the coming 5.x releases. So hop aboard, and come experience where this ride will take you! And as always, never hesitate to submit enhancement request or feedback.