Engage: Topical Text Analysis

Contents

Overview

Note: This feature is currently in beta. If you'd like to learn more about our beta program for this feature, please reach out to your Program Architect or Customer Success Manager. 

Our topic text analysis feature uses artificial intelligence (AI) to analyze comments and discover the topics that are most important to an organization's employees. This, combined with the sentiment analysis of those topics, help people leaders quickly understand what their employees feel is working and what can be improved. 

Survey Reports

Navigate to the survey you'd like to review and click "See Report". From there, you'll have access to several sections. However, the sections that pertain to the topic text analysis feature are Topics, Comments, Categories, and Questions.

Topics

Bubble Chart

bubble_chart.png

  • The view will display up to 10 topics.
  • The size of the bubble is proportional to the # of comments that reference the topic.
  • The # of comments that reference the topic will display in the bubble (under the topic name).
  • The topic order is as follows: positive on the left, neutral in the middle, and negative on the right.
  • Hovering over each bubble will trigger a modal outlining the distribution of positive, neutral, and negative sentiments. This will also be mirrored in the colors around the bubble where the amount of each color indicates the amount of that sentiment (i.e. blue = positive, gray = neutral, and red = negative).

Word Cloud

word_cloud.png

  • The view will display up to 10 topics.
  • The size of the word is proportional to the # of comments that reference the topic.
  • Hovering over each word will trigger a modal showing the distribution of positive, neutral, and negative sentiments. This will also be mirrored in the color of the word where the color indicates the average sentiment (i.e. blue = positive, gray = neutral, and pink/red = negative). 

Drill Down

Below either of these views is a drill down that shows the:

  • Topic name.
  • # of questions where the topic was referenced in the comment responses.
  • # of question categories where the topic was referenced in the comment responses.
  • # of comments where the topic was referenced.
  • Sentiment distribution of the topic.

drill_down.png

The topic details provide an opportunity to segment the data further by department or other dimensions:

dimensions.png

segment.png

You can also see the specific comments that reference the topic:

comments.png

Comments

This section shows the comments in the survey along with the question, topic and sentiment of each:

comments_section.png

Categories

This section shows a small preview of the topics at a category level (open text categories only).

categories_section.png

Questions

This section shows a small preview of the topics (open text questions only). 

questions_section.png

FAQs

How does the topical text analysis feature select topics? 

There is a predetermined list of topics associated with specific keywords. Each comment is scanned for those keywords. 

What if a comment doesn't contain any keywords?

The comment will be mapped to the "Miscellaneous" topic. As the AI learns and the topic list evolves, this will be less common. 

What if a comment contains keywords for more than one topic?

The comment will be analyzed further to determine which topic takes up the majority of the comment. The comment will then be mapped to that topic. 

Can topical text analysis be used on comments that are in a language other than English? 

Yes. The feature uses the Amazon Web Services (AWS) API for translations and will automatically translate the comment into English, then scan it for keywords. The support languages can be found here