Article-Summarizer-Using-AI

An AI-based web application that provides concise summaries of articles using advanced natural language processing (NLP) techniques.

Table of Contents

• Introduction
• Data Exploration
• Model Selection
• Model Fine-Tuning
• Extractive Summarization
• Web Application Development
• Installation
• Usage

Introduction

Article-Summarizer-Using-AI is a web application designed to summarize lengthy articles using NLP. The application allows users to upload their own articles or use sample data to generate summaries in various styles, utilizing a generative AI model.

Data Exploration

Dataset

The dataset used for training and evaluation is the PubMed Summarization dataset. It includes articles from PubMed with corresponding abstracts used as summaries.

1. Loading the Dataset:

   from datasets import load_dataset
   
   pubmed_data = load_dataset("ccdv/pubmed-summarization", split='train[:1000]')

2. Initial Data Cleaning:

   • Remove rows with missing values to ensure data quality.

   pubmed_data = pubmed_data.filter(lambda x: x['article'] is not None and x['abstract'] is not None)

3. Exploratory Data Analysis:

   • Examine the distribution of article lengths and summary lengths.
   • Identify common topics and terminology within the dataset.

   print(pubmed_data[0])  # View the first data entry

Model Selection

Preprocessing

1. Text Tokenization:

   • Split text into sentences and words for detailed analysis.

   from nltk.tokenize import sent_tokenize, word_tokenize
   
   sentences = sent_tokenize(article_text)
   words = word_tokenize(sentence)

2. Stop Words Removal:

   • Remove common English words that do not contribute to the summary.

   from nltk.corpus import stopwords
   
   stop_words = set(stopwords.words('english'))
   words = [word for word in words if word.lower() not in stop_words]

3. Lemmatization:

   • Convert words to their base forms.

   from nltk.stem import WordNetLemmatizer
   
   lemmatizer = WordNetLemmatizer()
   words = [lemmatizer.lemmatize(word.lower()) for word in words]

Generative Model

1. API Configuration:

   • Use the google.generativeai library for model generation.

   import google.generativeai as genai
   import os
   
   api_key = os.environ.get('your_api_key')
   genai.configure(api_key=api_key)

2. Model Initialization:

   • Set up the generative AI model.

   model = genai.GenerativeModel()

Model Fine-Tuning

Training

• Fine-tune the model with the PubMed dataset to improve summary quality.

  # Example pseudo-code for fine-tuning
  model.train(dataset=pubmed_data, epochs=10, learning_rate=0.001)

Extractive Summarization

Approach

For extractive summarization, the application uses traditional NLP techniques to identify key sentences from the article without relying on a generative model.

1. Extractive Summary Script:

   Rename the provided extractive_summary.py to app.py and move it to the project root:

   mv /mnt/data/extractive_summary.py app.py

2. Core Logic:

   • The extractive summarization script uses statistical and heuristic methods to identify the most important sentences in the text.

   # Example of extractive summarization
   def extractive_summary(text):
       # Tokenize the text and rank sentences
       sentences = sent_tokenize(text)
       # Rank and select key sentences (pseudo-code)
       summary = ' '.join(sentences[:3])  # Example: Select first 3 sentences
       return summary

3. Integration:

   • Integrate the extractive summarization logic with the Flask application to allow users to choose between generative and extractive summaries.

   @app.route('/summarize', methods=['POST'])
   def summarize():
       if 'file' in request.files and request.files['file'].filename != '':
           file = request.files['file']
           article_text = file.read().decode("utf-8")
       else:
           sample_index = int(request.form['sample'])
           article_text = pubmed_data[sample_index]['article']
   
       style = request.form.get('style', 'brief')
       summary_method = request.form.get('method', 'generative')
       
       if summary_method == 'generative':
           summary_text = preprocess_and_summarize(article_text, style)
       else:
           summary_text = extractive_summary(article_text)
   
       return render_template('result.html', original=article_text, summary=summary_text)

Evaluation

• Evaluate the model's performance using metrics such as ROUGE or BLEU.

  from nltk.translate.bleu_score import sentence_bleu
  
  reference = [reference_summary.split()]
  candidate = generated_summary.split()
  score = sentence_bleu(reference, candidate)
  print(f'BLEU Score: {score}')

Web Application Development

Backend

• Flask Setup:

  • Initialize the Flask app and configure the login manager.

  from flask import Flask
  from flask_login import LoginManager
  
  app = Flask(__name__)
  app.secret_key = 'your_secret_key'
  login_manager = LoginManager(app)

• Routes and Authentication:

  • Implement routes for login, registration, summarization, and logout.

  @app.route('/login', methods=['GET', 'POST'])
  def login():
      # login logic here
      return render_template('login.html')

Frontend

• Templates:

  • Create HTML templates for the user interface.

  <!-- templates/index.html -->
  <form action="{{ url_for('summarize') }}" method="post" enctype="multipart/form-data">
      <input type="file" name="file">
      <button type="submit">Summarize</button>
  </form>

• User Experience:

  • Ensure a user-friendly interface with clear instructions and feedback.

Installation

Prerequisites

• Python 3.7+
• Flask
• NLTK
• Generative AI Library (e.g., google.generativeai)
• An API key for generative AI

Steps

1. Clone the Repository:

   git clone https://github.com/yourusername/Article-Summarizer-Using-AI.git

2. Navigate to the Project Directory:

   cd Article-Summarizer-Using-AI

3. Create a Virtual Environment:

   python -m venv venv
   source venv/bin/activate  # On Windows use `venv\Scripts\activate`

4. Install Dependencies:

   pip install -r requirements.txt

5. Set Environment Variables:

   • Create a .env file with your API key.

   your_api_key=<YOUR_GENERATIVE_AI_API_KEY>
   

6. Download NLTK Data:

   The script handles downloading necessary NLTK data.

Usage

1. Run the Application:

   flask run --port=5001

2. Access the App:

   • Visit http://127.0.0.1:5001 in your browser.

3. Login/Register:

   • Register a new account or log in with existing credentials.

4. Summarize Articles:

   • Upload a text file or choose a sample to summarize.

5. View Summary:

   • The summarized text is displayed on the results page.

Thank you for using Article-Summarizer-Using-AI! We hope you find it useful for your summarization needs.