![Methodology
• Data Collection Methods:
• Data Gathering: Collect 3 book-related dataset, including- Books: ([‘Book-Author', 'Year-Of-Publication',
'Publisher’, 'Image-URL-S', 'Image-URL-M', 'Image-URL-L']), Users: (['User-ID', 'Location', 'Age']), Ratings:
(['User-ID', 'ISBN','Book-Rating']), from Kaggle.
• Preprocessing: we choose those user-id’s who give ratings >100 and choose those books whose ratings are >30.
Then we use Train_test_spit to train our models.
• Data Analysis Techniques/Technologies:
• we use some regression algorithm to train the model.
• Regression Algorithms Training & Testing:
• Naive Bayes
• Decision Tree
• Random Forest
• k-Nearest Neighbors (KNN)
• XGBoost](https://image.slidesharecdn.com/presentationtemplategroup6-241121081015-fb6da7c7/85/PRESENTATION-TEMPLATE-OF-LIBROLINK-A-BOOK-5-320.jpg)
PRESENTATION TEMPLATE OF LIBROLINK A BOOK
- 1. LIBROLINK • GROUP NUMBER - 52 • NAME – RIDDHI PRATIM SHEE, SUBHAM GANGULY, SUBHAM MONDAL,TUHIN BISWAS • STREAM – B.TECH C.S.E (AI & ML) • ENROLLMENT – 2111200010013, 2111200010022, 2111200010033, 2111200010035 • REGISTRATION -210100132892, 210100220312, 210100335959, 210100354818 • DEPARTMENT - COMPUTER SCIENCE ENGINEERING • NAME OF SUPERVISOR with AFFILIATION – Dr. Soumi Dutta
- 2. INTRODUCTION • OVERVIEW OF PROJECT / BACKGROUND AND CONTEXT • Our project focuses on developing an intuitive Book Recommendation System tailored to today's digital landscape. By leveraging algorithms and data analytics, we aim to provide personalized book suggestions by analyzing user preferences and book metadata. The goal is to enhance the book discovery process with a user-friendly interface. • IMPORTANCE OF THE PROJECT • Enhanced User Experience: Tailors book suggestions to individual tastes, reducing search time. • Increased Engagement: Boosts user interaction by consistently recommending relevant content. • Data-Driven Insights: Provides valuable information on reading trends and behaviors for strategic decisions.
- 3. PROBLEM STATEMENT • Description of the Problem Develop a machine learning-based book recommendation system to provide personalized recommendations to users based on their searching preferences. The system should aim to improve user satisfaction, engagement, and retention by accurately predicting books that users are likely to enjoy, thus enhancing their overall reading experience. • Objectives and Goals • Personalization: Tailor recommendations to individual preferences for enhanced user satisfaction. • Increased Engagement: Encourage more time spent on the platform with compelling suggestions. • Discovery: Introduce users to new authors and genres for broader reading experiences. • Continuous Improvement: Collect and analyze data to refine recommendations over time.
- 4. LITERATURE REVIEW Summary of Relevant Research • Collaborative Filtering: Utilizes user interaction data to find and recommend books based on similarities between users or items. • Content-based Filtering: Recommends books by analyzing user preferences and book attributes. • Hybrid Methods: Combines collaborative and content-based filtering for more accurate and diverse recommendations. • Matrix Factorization: Decomposes user-item interaction matrices to uncover latent factors influencing book preferences. Gaps in Existing Knowledge or Applications • Cold Start Problem: Difficulty in recommending for new users or books with little interaction data. • Scalability Issues: Challenges in maintaining performance and accuracy with growing data volumes. • Dynamic Preferences: Insufficient adaptation to users' changing tastes over time.
- 5. Methodology • Data Collection Methods: • Data Gathering: Collect 3 book-related dataset, including- Books: ([‘Book-Author', 'Year-Of-Publication', 'Publisher’, 'Image-URL-S', 'Image-URL-M', 'Image-URL-L']), Users: (['User-ID', 'Location', 'Age']), Ratings: (['User-ID', 'ISBN','Book-Rating']), from Kaggle. • Preprocessing: we choose those user-id’s who give ratings >100 and choose those books whose ratings are >30. Then we use Train_test_spit to train our models. • Data Analysis Techniques/Technologies: • we use some regression algorithm to train the model. • Regression Algorithms Training & Testing: • Naive Bayes • Decision Tree • Random Forest • k-Nearest Neighbors (KNN) • XGBoost
- 6. Methodology • Model Evaluation : • Calculate Mean Squared Error (MSE) for each algorithm to assess their predictive performance. • Selection of Optimal Algorithm: • Identify XGBoost as the most suitable algorithm based on the lowest MSE score obtained during testing. And utilize XGBoost as the primary model for our book recommendation system due to its superior performance. • Deployment: • Integrate the trained XGBoost model into the book recommendation system using streamlit for frontend, allowing users to receive personalized recommendations based on their search.
- 7. Outcome • Results : Naïve Bayes XGBoost Decision Tree Random Forest KNN 17.656 19.753 12.87 13.469 12.332 Baseline MSE – 13.134 Optimal
- 8. Outcome • Potential Impact • Enhanced Discovery: Recommendation system can help users discover new books, genres, or authors they may not have found otherwise, broadening their literary horizons. • Increased Engagement: With better recommendations, user are more likely to engage with books and reading platforms.
- 9. Significance Practical Applications • Digital Libraries and Reading Apps: Offer tailored book suggestions to users to facilitate content discovery and engagement. • Educational Platforms: Provide students with supplementary reading materials based on course topics and learning preferences. • Public Libraries: Assist patrons in discovering new books and authors based on their reading history and interests. • Contribution to the Field • Algorithmic Advancements: Selecting XGBoost enhances recommendation accuracy and relevance. • Performance Benchmarking: Establishing a standard by comparing various classification algorithms. • User-Centric Focus: Prioritizing accuracy for improved user satisfaction and engagement.
- 10. Timeline • Project Phases: • 6 months • Milestones and Deadlines: • We will update our frontend part using react and we will use ensemble learning and unsupervised learning to get more optimized result. We will create a user friendly interface where when user click on thumbnail, user will get all the details of the book (i.e. author name , rating , price) as well as a synapsis of the book
- 11. Conclusion In conclusion, our presentation has shed light on the crucial aspects of book recommendation systems, emphasizing the importance of algorithm selection, and user-centric design. Through our evaluation, XGBoost emerged as a standout algorithm, promising enhanced accuracy and relevance in book recommendations • Summary of Key Points: • Algorithmic Advancements: XGBoost was identified as the optimal algorithm for improving recommendation accuracy and relevance. • User-Centric Focus: Prioritizing accuracy enhances user satisfaction and engagement with recommendation systems. • Future Directions: Looking ahead, there are several avenues for future exploration and innovation in the field of book recommendation systems. These include: • Incorporating Contextual Information: Integrating user context such as location, time, and social interactions for more personalized recommendations. • Addressing Diversity and Fairness: Ensuring recommendations reflect diverse perspectives.
- 12. THANK YOU