NLP - US Patent Phrase to Phrase Matching

This is a NLP project focuses on analyzing and matching phrases from US patent documents from Kaggle Competition.

Installation

1. Clone the repository

git clone https://github.com/waijian1/nlp_phrase_match.git
cd nlp_phrase_match

2. Create and activate conda environment

conda env create -f environment.yaml
conda activate nlp_phrase_match # replace with your actual environment name

Project Structure

├── main.ipynb          # Main notebook containing analysis and results
├── environment.yaml    # Conda & pip packages environment file
└── us-patent-phrase-to-phrase-matching/    # Data directory (download from notebook)

Notebook Overview

You can view the complete notebook with all outputs and visualizations in these formats:

• View in nbviewer (recommended)
• View in GitHub

The notebook includes:

• Download train & validation data from Kaggle competition
• Data preprocessing
• Fine-tune pretrained Transformer model from HuggingFace
• Model training
• Results

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.