How is sentiment calculated in the Text Analysis Report?
Sentiment is assigned to each comment your survey recieves. To assign a sentiment, our system measures how positive or negative the comment is. We then display the sentiment with the highest score.
Each comment will be assigned one of the following:
- positive
- negative
- neutral
- mixed.
Free-text questions
For free-text questions, our system evaluates the comment in the context of the question.
Example #1:
Question: Do you feel like you’ve been bullied at work?
Comment: Yes, I have.
| Positive | Neutral | Negative | Mixed | Assigned Sentiment |
| 0.1 | 0.1 | 0.8 | 0.0 | Negative |
This is an example of a poorly written free-text question where a positive answer has a negative sentiment. However, as we evaluate the comment in the context of the question, our system correctly assigns it as Negative.
Let's look at the same question with the opposite answer.
Example #2:
Question: Do you feel like you’ve been bullied at work?
Comment: No, I have not.
| Positive | Neutral | Negative | Mixed | Assigned Sentiment |
| 0.0 | 1.0 | 0.0 | 0.0 | Neutral |
Questions followed by a comment
Our system also evaluates the wider context if participants comment after answering a question. In this situation, we look at the original question asked, their answer and how their comment relates to that.
Example #1:
Question: [ORGANISATION_NAME] is a great place to work
Answer: Agree
Comment: I really like working at [ORGANISATION_NAME] because lunchtime conversations are fun and I get recognition through Bravos for the work I do.
| Positive | Neutral | Negative | Mixed | Assigned Sentiment |
| 0.9 | 0.1 | 0.0 | 0.0 | Positive |
Why the sentiment might be incorrect
When we automatically assign sentiment, it is a "best guess". This is because language can be hard for computers to understand.
For example:
- Languages are used by billions of people, and all of them use it in a different way.
- Computers can only recognise patterns and try to respond based on those patterns.
- Context is essential to interpreting language and there might not always be enough context in a comment.
That means there may be times you disagree with the sentiment we display for a comment. If this happens, you might want to change the sentiment.
How we protect your data when analysing comments
At AskYourTeam, we understand the importance of protecting your personally identifiable information (PII) and take every precaution to ensure it's safe and secure.
Sentiment is automatically assigned to comments using AI. So we redact all personally identifiable information (like names) before it's sent to the third-party AI system to be analysed and then replaced as the data is returned to AskYourTeam's secure systems. Your identity and data is secure and protected at all times.
Learn more about how we protect your data when using AI.
Why doesn't my Text Analysis Report show sentiment?
If you can't see sentiment in your Text Analysis Report, it might be because:
- You need to enable AI for reporting in your survey settings.
- Your organisation has opted out of AI features.
Can sentiment be assigned to languages other than English?
At the moment, sentiment can only be automatically calculated and assigned for comments written in English. We are working to support more languages in the future.
If comments are given in other languages, they will appear as 'neutral'.
If you have purchased the comment translation upgrade, the comment will be translated into English and then the sentiment will be assigned.