@November 2, 2022
“🟢 Amazing! 🚀”, “🟡 Good start but could be better. 🤷♀️”, “🔴 Not very nice. 😬”
are all things your might have seen your community members say. From now on, though, you need not worry about not knowing what emotional tone their messages carry, as…
…we just released the widely requested Content: Sentiment feature!
👩🏽🏭 How does it work?
- we pass the contents of your member activities to the Amazon Comprehend API, which returns their sentiment and confidence score
- if you think the automatically assigned sentiment is not up to snuff, we allow for manual overwrite of the sentiment value
- we also plot charts of
Overall Sentiment
andSentiment Over Time
for keywords to help you understand sentiment trends - you can also use the new
Activity Sentiment filter
to hone in on a specific sentiment subset (for example, just activities with negative sentiment to understand what your community members dislike) - the technical nitty-gritty
- we sentiment-scored 30days worth of activities for our users + any new activity that will come in
- the activity types we score are listed in this doc
‣
threshold | condition |
SLIGHTLY POSITIVE | if positive over 50% |
VERY POSITIVE | if positive over 65% |
OVERWHELMINGLY POSITIVE | if positive over 80% |
SLIGHTLY NEGATIVE | if negative over 50% |
VERY NEGATIVE | if negative over 65% |
OVERWHELMINGLY NEGATIVE | if negative over 80% |
📽️ If video is more your jam, check out this Loom demo!
🧙🏽♀️ Why is it awesome?
- because you can now understand the emotional tone of your community contributions
- observe trends such as “Is my negative sentiment ratio decreasing?”
- in tandem with our
keywords
feature, evaluate, for example, the reception of new features
👩🏿🍳 What are we cooking up next?
- as mentioned previously, AI keyword suggestions are not far off, wrapping up the holy trinity of our Content features!
🚦 Now the only thing left is to understand the sentiment of your community. Try it out in your account!