---
title: SenseCatch
emoji: 🎭
colorFrom: blue
colorTo: indigo
sdk: docker
app_port: 7860
pinned: false
license: mit
short_description: Sentiment analysis with seven selectable models
---

# SenseCatch

SenseCatch labels English review text as Positive or Negative. It serves seven selectable models from a single container: four classical machine learning models (Naive Bayes, Logistic Regression, LinearSVC, NBSVM), a DistilBERT transformer fine-tuned on IMDB, a stacked ensemble that combines them, and an optional rule-based mode.

The first request after the Space has been idle is slow, because the container wakes from sleep and loads the models on demand. Later requests are fast.

Source code: https://github.com/berkinksk/SenseCatch