Sentiment analysis

Sentiment analysis that knows when praise is sarcasm

Most sentiment tools were built for English and translated. Qalu reads Arabic as people actually write it: dialects, shorthand, jokes and sarcasm that reads like praise.

  • News and 8 platformsFacebook, X, Instagram, TikTok, YouTube, LinkedIn, Threads and Reddit
  • Arabic and its dialectsEgyptian, Gulf, Levantine, Iraqi and Maghrebi, plus English
  • CanadaData is kept in Montreal; Qalu is made by InfoTimes

What we label

  • Sentiment toward your organisation: positive, neutral, negative or mixed.
  • Stance, which is not sentiment: a negative story about a sector may not oppose you.
  • Emotions: anger, trust, fear, sadness, joy, surprise and disgust.
  • Sarcasm, counted as negative even when the words sound kind.

Why you can trust it

  • Every label comes with the phrases it rests on.
  • Correct any label yourself from the review page.
  • An accuracy card: how often human review agrees with the labels in your project.

Sentiment FAQ

Which dialects are covered?

Egyptian, Gulf, Levantine, Iraqi and Maghrebi Arabic, Modern Standard Arabic and English.

Test the labels on your data

Send us a sample of comments or posts and we show you how they are read.