[Project Addition]: AI Based Book Recommendation System

[Sanjeev-Kumar78]

Deep Learning Simplified Repository (Proposing new Project)

:red_circle: Project Title : Book Recommendation AI
:red_circle: Aim : To develop an AI-based book recommendation system that suggests books of similar type or genre based on the book name provided by the user.
:red_circle: Dataset : GoodReads-100K

:red_circle: Approach :

1. Data Preprocessing:

   • Load the Goodreads Books dataset from Kaggle.
   • Clean and preprocess the data, focusing on book titles, genres, and descriptions.
   • Use tf-idf vectorization to convert textual information into numerical format.

2. Initial Model Training:

   • Train a baseline recommendation model using tf-idf vectors and cosine similarity to suggest books based on title or genre similarity.
   • Evaluate the performance of this baseline model to establish a reference point.

3. AI Model Integration:

   • Select a suitable pre-trained transformer model from Hugging Face, such as BERT or DistilBERT, known for its capability in handling text data.
   • Fine-tune the selected Hugging Face model on the preprocessed Goodreads dataset to capture deeper semantic relationships between book titles and descriptions.
   • Save and document the fine-tuned model for further use.

4. Combining Models:

   • Develop a hybrid recommendation system by combining the tf-idf based model and the fine-tuned transformer model.
   • Use the tf-idf model for initial filtering of recommendations and the transformer model for refining and ranking the recommendations.
   • Implement a mechanism to fetch additional book data from external sources on the internet, enhancing the recommendation system's database.

5. Evaluation and Optimization:

   • Test the combined model's performance using various metrics such as precision, recall, and F1 score.
   • Optimize the model based on feedback and evaluation results to improve accuracy and relevance of recommendations.

6. Deployment and User Interface:

   • Create a user-friendly interface where users can input book titles to receive recommendations.
   • Ensure the system is scalable and can handle multiple users simultaneously.
   • Deploy the model on a suitable platform, providing access to a wide range of users.

---

📍 Follow the Guidelines to Contribute in the Project :

• You need to create a separate folder named as the Project Title.
• Inside that folder, there will be four main components.
  • Images - To store the required images.
  • Dataset - To store the dataset or, information/source about the dataset.
  • Model - To store the machine learning model you've created using the dataset.
  • requirements.txt - This file will contain the required packages/libraries to run the project in other machines.
• Inside the Model folder, the README.md file must be filled up properly, with proper visualizations and conclusions.

---

:red_circle::yellow_circle: Points to Note :

• The issues will be assigned on a first come first serve basis, 1 Issue == 1 PR.
• "Issue Title" and "PR Title should be the same. Include issue number along with it.
• Follow Contributing Guidelines & Code of Conduct before start Contributing.

---

:white_check_mark: To be Mentioned while taking the issue :

• Full name : Sanjeev Kumar
• GitHub Profile Link : https://www.github.com/Sanjeev-Kumar78
• Email ID :sanjeev-kumar78@outlook.com
• Participant ID (if applicable):
• Approach for this Project : Mentioned above
• What is your participant role?: GSSOC'24 Contributor Role

---

Happy Contributing 🚀

All the best. Enjoy your open source journey ahead. 😎

[abhisheks008]

Hi @Sanjeev-Kumar78 thanks for showing up with this idea. Need to execute at least 3 deep learning models for this dataset/project. Compare them all and find the best fitted one and conclude the same based on their accuracy scores.

Please share your updated approach on the same.

[khushi-igupta]

Hello I am a GsSOC'24 contributor and I would like to work on this issue. Can you please assign it to me?

[Sanjeev-Kumar78]

Updated Approach:

1. Data Preprocessing:

• Load the Goodreads Books dataset from Kaggle.
• Clean and preprocess the data, focusing on book titles, genres, and descriptions.
• Use tf-idf vectorization to convert textual information into numerical format for baseline comparison.

2. Baseline Model:

• Model 1: TF-IDF Vectorization with Cosine Similarity

  • Train a baseline recommendation model using tf-idf vectors and cosine similarity to suggest books based on title or genre similarity.
  • Evaluate the performance of this baseline model to establish a reference point.

3. Deep Learning Models:

• Model 2: Neural Collaborative Filtering (NCF)

  • Implement NCF using user-book interaction data, incorporating both user and item embeddings.
  • Train the model to predict user-book interactions and recommend books based on these predictions.
  • Evaluate the model's performance using metrics like precision, recall, and F1 score.

• Model 3: Convolutional Neural Network (CNN) for Text Data

  • Use CNN to process book titles and descriptions, capturing local and sequential patterns.
  • Train the model on the preprocessed dataset, focusing on text data to generate recommendations.
  • Evaluate the CNN model's performance and compare it with other models.

• Model 4: Transformer-based Model from Hugging Face

  • Fine-tune a pre-trained transformer model (e.g., BERT or DistilBERT) on the Goodreads dataset.
  • Use the fine-tuned model to capture deep semantic relationships between book titles and descriptions.
  • Evaluate the transformer model's performance and compare it with other models.

• Combining Models and Hybrid Approach:

  • Develop a hybrid recommendation system by combining the strengths of the tf-idf, NCF, CNN, and transformer models.
  • Use the tf-idf model for initial filtering of recommendations, then refine and rank the recommendations using deep learning models.

• Evaluation and Comparison:

  • Test all models' performance using various metrics such as precision, recall, F1 score, and Mean Squared Error (MSE).
  • Compare the accuracy and performance of each model to determine the best fitted one.
  • Document the evaluation results and draw conclusions based on the comparative analysis.

  ---

  @abhisheks008 is it possible that @Sanjeev-Kumar78 & @khushi-igupta can work together on this project.

[abhisheks008]

No @Sanjeev-Kumar78 one contributor to be assigned for one issue only. As you have raised this issue you can work on this only. Should I assign this issue to you?

@khushi-igupta you can check out other open issues present here in this repo.

[Sanjeev-Kumar78]

Okay, Assign me @abhisheks008

[abhisheks008]

Assigned @Sanjeev-Kumar78

[khushi-igupta]

Okay

On Wed, May 22, 2024, 10:04 Abhishek Sharma @.***> wrote:

No @Sanjeev-Kumar78 https://github.com/Sanjeev-Kumar78 one contributor to be assigned for one issue only. As you have raised this issue you can work on this only. Should I assign this issue to you?

@khushi-igupta https://github.com/khushi-igupta you can check out other open issues present here in this repo.

— Reply to this email directly, view it on GitHub https://github.com/abhisheks008/DL-Simplified/issues/614#issuecomment-2123847969, or unsubscribe https://github.com/notifications/unsubscribe-auth/AZMQA6MPWQP3ZCOYYEXXBZ3ZDQN5VAVCNFSM6AAAAABIBO3YFWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMRTHA2DOOJWHE . You are receiving this because you were mentioned.Message ID: @.***>

[github-actions[bot]]

Hello @Sanjeev-Kumar78! Your issue #614 has been closed. Thank you for your contribution!