Which levels does sentiment analysis distinguish?

Three: positive, neutral and negative. Neutral stands for factual or mixed statements without a clear judgement. Alongside the sentiment for each topic, deepsight also classifies the overall impression of a response; if a response praises one topic and criticizes another, the classification for each topic tells you more.

Does sentiment analysis recognize irony?

Not reliably. The classification takes the context of the response into account, not just individual words. Irony often remains ambiguous, though, and there is no separate level for unclear cases: an ironic response is also classified as positive, neutral or negative.

Can I compare sentiment between groups?

Yes. With group comparison, you see how sentiment differs between groups, such as by region or customer type. The attributes for this come from your data.

Are responses in other languages classified too?

Yes. Analyze responses in other languages, too: an integrated translation (AI-powered, or with DeepL or Google if you prefer) supports the analysis.

deepsight cloud · Sentiment analysis Which topics get praise, and which get criticism?

Look at sentiment in the context of each individual response.

Example

One response, several sentiments

Sentiment is classified for each topic, not just for the response as a whole. You check every classification against the original response.

Open-ended response · Restaurant visit

“The food was excellent, the service rather slow. We waited 40 minutes for a table – far too long.”

Select a category to see the matching passage.

Classification

Positive, neutral or negative for each topic in a response
In context: “slow” is criticism when it describes the service, but perhaps not when it describes a relaxed atmosphere
Irony is not always clear-cut: check conspicuous classifications against the original response

Illustrative example.

FAQ

Questions about sentiment analysis

You can compare praise and criticism per touchpoint – from booking to support – in a touchpoint analysis, for example.

More features · Analyze responses

Topic analysisOrganize open-ended responses by topic, with new categories or your own codebook.View Intent analysisSort comments by what people want, such as a complaint, praise, a question or an intention to cancel.View NPS analysisCompare topics between promoters, passives and detractors.View Group comparisonCompare topics and ratings by attributes from your data.View

See the feature with your own data

We will walk you through the analysis using an example from your field.

Request a demo