Skip to content

Latest commit

 

History

History
217 lines (170 loc) · 6.49 KB

Report.md

File metadata and controls

217 lines (170 loc) · 6.49 KB

Hotel Recommender System - Web Mining Project Documentation

This document explains the implementation of a hotel recommendation system that uses web mining techniques to scrape hotel data from Google Travel and provide personalized recommendations based on user preferences.

Table of Contents

  1. Project Overview
  2. Architecture
  3. Web Mining Implementation
  4. Frontend Implementation
  5. Recommendation Algorithm
  6. Technical Components

Project Overview

The Hotel Recommender System is a web application that:

  1. Takes user input for a city and preferred hotel features
  2. Scrapes real-time hotel data from Google Travel
  3. Processes the scraped data
  4. Generates personalized hotel recommendations using similarity matching
  5. Presents the results in an interactive interface

Architecture

The project follows a client-server architecture:

  • Backend: Python Flask server handling web scraping and recommendations
  • Frontend: HTML/CSS/JavaScript interface for user interaction
  • Key Libraries:
    • Selenium (web scraping)
    • Flask (server)
    • scikit-learn (similarity calculations)
    • pandas (data processing)

Web Mining Implementation

1. Web Scraping Setup

The project uses Selenium WebDriver to automate the scraping of Google Travel's hotel data. Here's the core scraping setup:

from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.common.action_chains import ActionChains
from selenium.webdriver.common.keys import Keys

def initialize_scraper():
    driver = webdriver.Chrome()
    base_url = 'https://www.google.com/travel/hotels'
    driver.get(base_url)
    return driver

2. Data Extraction Functions

The project includes several specialized functions to extract different hotel attributes:

def get_hotel_name(div):
    return div.find_element(By.CSS_SELECTOR, '.BgYkof.ogfYpf').text

def get_hotel_rating(div):
    try:
        return div.find_element(By.CSS_SELECTOR, '.lA0BZ').text
    except:
        return 'NA'

def get_hotel_price(div):
    try:
        return div.find_element(By.CSS_SELECTOR, '.kixHKb.flySGb').text
    except:
        return 'NA'

def get_hotel_features(div):
    try:
        features = div.find_elements(By.CSS_SELECTOR, '.bX73z')
        hotel_features = [feature.text for feature in features]
        while len(hotel_features) < 9:
            hotel_features.append('NA')
        return hotel_features
    except:
        return ['NA'] * 9

3. Web Mining Process Flow

The web mining process follows these steps:

  1. Search Initialization:
search_bar = driver.find_element(By.CLASS_NAME, 'II2One')
search_bar.clear()
search_bar.send_keys(city)
time.sleep(5)  # Wait for suggestions
  1. Data Collection:
div_elements = driver.find_elements(By.CSS_SELECTOR, '.kCsInf')
hotels_data = []

for div in div_elements:
    hotel_data = {
        'Hotel_Name': get_hotel_name(div),
        'Hotel_Rating': get_hotel_rating(div),
        'Hotel_Price': get_hotel_price(div),
        'City': city,
        'URL': get_hotel_url(div)
    }
    
    # Convert features to binary format
    for feature in selected_features:
        hotel_data[feature] = 1 if feature in features else 0
        
    hotels_data.append(hotel_data)

Recommendation Algorithm

The system uses cosine similarity to match user preferences with hotel features:

from sklearn.metrics.pairwise import cosine_similarity

def generate_recommendations(hotels_df, user_features):
    # Calculate similarity scores
    similarity_scores = cosine_similarity([user_features], 
                                       hotels_df[selected_features])
    
    # Add similarity scores to DataFrame
    hotels_df['Similarity'] = similarity_scores[0]
    
    # Sort and get top recommendations
    recommendations = hotels_df.sort_values(by='Similarity', 
                                          ascending=False).head(10)
    return recommendations

Frontend Implementation

1. User Interface

The frontend provides an intuitive interface with:

  • City input field
  • Feature selection checkboxes
  • Results display area with hotel cards

2. Dynamic Data Processing

The frontend processes user selections into a binary feature vector:

const featureList = [
    "Free breakfast", "Free Wi-Fi", "Air conditioning",
    "Restaurant", "Free parking", "Room service",
    "Pool", "Full-service laundry", "Fitness centre",
    "Kitchen", "Airport shuttle", "Spa"
];

const binaryFeatures = featureList.map(feature => 
    selectedFeatures.includes(feature) ? 1 : 0
);

3. Results Display

Hotel recommendations are displayed as cards with:

  • Hotel name and basic information
  • Similarity score
  • Direct booking link
  • Visual feedback during loading
let resultsHTML = '<h2>Recommended Hotels</h2><div class="hotel-list">';
data.forEach(hotel => {
    resultsHTML += `
        <div class="hotel-card">
            <h3>${hotel.Hotel_Name}</h3>
            <p>City: ${hotel.City}</p>
            <p>Rating: ${hotel.Hotel_Rating}</p>
            <p>Price: ${hotel.Hotel_Price}</p>
            <p>Similarity Score: ${(hotel.Similarity * 100).toFixed(2)}%</p>
            <a href="${hotel.URL}" target="_blank" class="book-link">
                Book Now
            </a>
        </div>
    `;
});

Technical Components

Key Features:

  1. Real-time Data: The system scrapes live data rather than using a static database
  2. Error Handling: Robust error handling for web scraping and API responses
  3. Response Processing: Structured data extraction and transformation
  4. User Experience: Loading indicators and smooth animations
  5. Responsive Design: Grid-based layout that adapts to different screen sizes

Selected Features for Analysis:

selected_features = [
    'Free breakfast', 'Free Wi-Fi', 'Air conditioning',
    'Restaurant', 'Free parking', 'Room service',
    'Pool', 'Full-service laundry', 'Fitness centre',
    'Kitchen', 'Airport shuttle', 'Spa'
]

Conclusion

This web mining project demonstrates the integration of:

  • Automated web scraping
  • Real-time data processing
  • Machine learning-based recommendations
  • Interactive user interface

The system provides a practical example of how web mining techniques can be applied to create a useful recommendation system that helps users find hotels matching their preferences.