Text Summarization

Overview

This notebook provides a comprehensive solution for text summarization using the BART Large Language Model (LLM). It includes functionality for extracting text from PDF files, generating embeddings, performing Retrieval-Augmented Generation (RAG), fine-tuning BART for improved performance, and displaying the output using the ipywidgets library.

Features

• Text Extraction from PDF: Extracts text content from PDF files.
• Embeddings: Generates embeddings for the extracted text using a pre-trained model.
• BART Model for Summarization: Utilizes the BART model for generating summaries.
• Retrieval-Augmented Generation (RAG): Enhances summarization by incorporating relevant retrieved documents.
• Fine-Tuning BART: Customizes the BART model for improved summarization performance on specific datasets.
• **Evaluating the fine-tuned model.
• Interactive UI: Uses the ipywidgets library for an interactive user interface to display the summarization results.

Requirements

Before running the notebook, ensure you have the following dependencies installed:

Acknowledgments

Hugging Face for the BART model and transformers library. PyMuPDF for PDF text extraction. Sentence-Transformers for embedding generation.

---

README: Webpage and Text Summarization Tool

This project implements a summarization tool using a fine-tuned BART model for generating summaries of text and web content. The tool provides two functionalities: summarizing web pages by fetching their content via a URL and summarizing user-provided text.

Features

1. Webpage Summarization:

   • Fetches the content of a webpage using its URL.
   • Generates a concise summary using the fine-tuned BART model.

2. Text Summarization:

   • Accepts user-inputted text.
   • Generates a summary of the text using the fine-tuned BART model.

3. Interactive Widgets:

   • Utilizes ipywidgets for a user-friendly interface.
   • Includes buttons for summarization and clearing inputs.

Requirements

Python Libraries

The following Python libraries are required:

• transformers: For the BART model and tokenizer.
• ipywidgets: For interactive input and output.
• requests: For fetching webpage content.
• IPython.display: For displaying widgets.

Installation

To install the necessary libraries, run the following command:

pip install transformers ipywidgets requests

Usage

1. Webpage Summarization

• Enter the URL of a webpage into the provided text input.
• Click the "Fetch & Summarize" button to fetch and summarize the webpage content.
• View the generated summary in the output area.

2. Text Summarization

• Enter the text to be summarized into the text area.
• Click the "Summarize Text" button to generate a summary.
• Clear the text area by clicking the "Clear Text" button if needed.

Code Overview

Model Setup

model_name = "fine_tuned_bart"
tokenizer = BartTokenizer.from_pretrained(model_name)
model = BartForConditionalGeneration.from_pretrained(model_name)

Summarization Functionality

Webpage Fetching and Summarization

def fetch_text_from_url(url):
    try:
        response = requests.get(url)
        response.raise_for_status()
        return response.text
    except requests.exceptions.RequestException as e:
        return str(e)

def summarize_text(text):
    inputs = tokenizer.encode("summarize: " + text, return_tensors="pt", max_length=1024, truncation=True)
    summary_ids = model.generate(inputs, max_length=512, num_beams=4, early_stopping=True)
    summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
    return summary

Interactive Widgets

# Widgets for URL input
url_input = Text(placeholder='Enter URL of the webpage...')
button_fetch_summarize = Button(description="Fetch & Summarize", button_style='success', icon="link")
button_clear_url = Button(description="Clear URL", button_style='danger', icon="trash")

# Widgets for Text Input
input_text_area = Textarea(placeholder='Enter text here...')
button_summarize_text = Button(description="Summarize Text", button_style='success', icon="cogs")
button_clear_text = Button(description="Clear Text", button_style='danger', icon="trash")

# Output Area
output = Output()

Event Handlers

• Fetch and summarize URL:

def handle_fetch_summarize(b):
    output.clear_output()
    url = url_input.value.strip()
    if url:
        text = fetch_text_from_url(url)
        if isinstance(text, str):
            with output:
                print(f"Error fetching URL: {text}")
        else:
            summary = summarize_text(text)
            with output:
                print("Summary:", summary)
    else:
        with output:
            print("Please enter a valid URL.")

• Summarize text:

def handle_summarize_text(b):
    output.clear_output()
    text = input_text_area.value.strip()
    if text:
        summary = summarize_text(text)
        with output:
            print("Summary:", summary)

Future Enhancements

1. Add support for additional languages.
2. Enable file uploads for batch text summarization.
3. Enhance error handling and user feedback.

---