Movie review sentiment

Write or paste an English movie review. A sentence-transformer (all-mpnet-base-v2) turns it into a 768-number embedding, and a classifier trained on 50,000 IMDB reviews decides whether it is positive or negative. Everything runs in your browser, so the text is not sent anywhere.

Loading…
How accurate is it?

Limitations:

  • Trained on full-length IMDB reviews. Very short texts, tweets and non-movie reviews are less reliable.
  • Only two classes: neutral or genuinely mixed reviews are forced into positive or negative, usually with a probability near 50 %.
  • Sarcasm and irony are often missed.
  • English only.

From the study Deep learning approaches for sentiment analysis of engaged audience in motion picture realm (Abir Hosen; supervisor Sultana Jahan Soheli, NSTU) · Embedding model: all-mpnet-base-v2 · Classifier: Hugging Face · Data: IMDB 50K movie reviews (Maas et al., 2011)