NLP Based Legal Text Summarization

Summarize Long Terms and Conditions Documents

SUMMARLY-LEGAL summarizes long Terms and Conditions documents using trained local Transformer models and a TF-IDF baseline.

Proposal-Based Models

TF-IDF Baseline

Extractive summarization baseline that selects important sentences based on word weights.

BART Trained Model

Main abstractive Transformer model loaded from local training checkpoint.

T5 and LED

Comparison models used to evaluate summarization performance against BART and baseline.

Summarizer Application

Upload or Paste Your Document

Choose one summarization method to summarize your Terms and Conditions document.

Input Document

BART is the main trained model used for abstractive summarization.

First run may take longer because the selected model checkpoint needs to be loaded.

Summary Result

0

Original Words

0

Summary Words

Detected Language: -

Used Model: -

About Project

How This Application Works

Training Checkpoint

The Transformer models are loaded from local trained model folders created during the training process.

TF-IDF Baseline

TF-IDF is used as the extractive baseline to compare against Transformer-based abstractive summarization models.

Document Input

Users can paste text directly or upload PDF and TXT files. The extracted text is cleaned before summarization.

Deployment Interface

The web application acts as an inference interface for the trained summarization models.