Sentiment Analysis automatically categorizes your survey text responses to reveal the emotion behind what people are saying.
You can filter survey comments by sentiments or scores by clicking into the survey and navigating to Comments. The sentiments are defined as:
- 'Positive' includes comments with a positive sentiment or a score of 9-10
- 'Neutral' includes scores 7-8
- 'Negative' includes negative sentiment or scores 0-6.
Visualizing the general sentiment of a survey can provide powerful and understandable insight while presenting a useful way to communicate results to other stakeholders.
The visualization represents comments based on comment topic, verbal sentiment, and quantity. You can find out more on how to interpret survey results here.
Question category/Automatic Filtering
The bubbles in the visualization represent the different topics covered in the survey comments. Comments in the survey are analyzed to identify common themes and sentiments. Sometimes, even if a category wasn't explicitly mentioned, a bubble can still be created for it if related keywords or topics are present. This helps to provide a more comprehensive view of the sentiment and related topics, and ensures all relevant insights are captured.
When you click on one of the bubbles, the view automatically filters to show comments in that topic.
Size/Quantity
The size of each bubble represents the share of comments relating to a topic in relation to others. For example, seeing the above visualization would mean your employees comment on manager support-related questions more than other topics.
Color/Sentiment
Sentiment analysis of commentary aggregates the general sentiment of a comment based on the wording. For example, 'I am happy' would be categorized as having a positive sentiment. This is represented in the visualization through color and location. Looking at the above image, this would mean that employees consistently highlight Manager Support as a positive aspect of the culture.
Sentiment analysis is operated by AWS Comprehend. For more information on how this works, please follow the link here.