README.md

---
title: Collocation Analyzer
emoji: 🔗
colorFrom: indigo
colorTo: purple
sdk: gradio
python_version: "3.10"
app_file: app.py
pinned: false
license: mit
---

# Interactive Collocation Analyzer

This is a premium, lightweight, interactive Collocation and Phrase-Pair extraction tool designed specifically for computational social science and humanities courses.

## Features
- **Flexible Data Input**: Upload datasets (`.csv`, `.xlsx`, or `.txt`).
- **N-Gram Choices**: Select between Bigrams (2-word pairs) or Trigrams (3-word triplets).
- **Association Metrics**:
  - **Raw Joint Frequency**: Ranks co-occurrences based purely on their combined count in the text.
  - **Pointwise Mutual Information (PMI)**: Highlights highly unique, semantic word pairings (such as idioms) even if they appear rarely, by dividing joint probabilities by individual word probabilities.
- **Plotly Visualizations**: Displays clean, responsive horizontal bar charts representing phrases ranked by score.
- **CSV Data Export**: Easily export extracted collocations and scores as a standard CSV report.