What is unique about Appbot Sentiment?
Appbot offers a proprietary Sentiment tool, designed to understand the nuanced way that language is used in reviews. We’ve built a new AI from the ground up and trained it on almost half a billion records. That means millions of records from every language that appears on the stores were used to train our models.
Our model is therefore much more accurate than off-the-shelf sentiment libraries, even on languages other than English. Speaking of languages; Appbot supports all languages! We calculate sentiment from the original language of the review, rather than from the English translation.
Appbot sentiment analysis has been designed to work equally well with app reviews and Amazon product reviews, as well as other types of short-format user feedback that our customers import via CSV or our API.
How accurate is Appbot’s sentiment analysis?
Our internal studies have reported accuracy of up to 93% – making this tool world-class in it’s accuracy.
How Appbot’s new Sentiment works
All Appbot accounts now show the new sentiment results. On the Reviews page, the new results look like this:
This replaces the old scale from Delighted to Angry. It’s more robust, especially for the different data sources we now support.
What each Appbot Sentiment category means
Here’s a brief description of each category of sentiment:
- Positive Sentiment: The comment is mostly positive.
- Negative Sentiment: The comment is mostly negative.
- Neutral Sentiment: The comment lacks any strong sentiment.
- Mixed Sentiment: The comment has mixed or conflicting sentiment.
If you have any questions about the new Appbot sentiment analysis or custom data sources, feel free to contact us