Book-Genre-Prediction

A Natural Language Processing (NLP) Project for predicting genres of a book using its summary as a feature.

The web app was hosted in a VM on Azure Cloud.

Description of File Structure:

Datasets:

This folder has all the datasets used for different experiments.

• test.xml and train.xml are used for Multi Label Genre Prediction.
• booksummaries.txt is used for Single Label Genre Prediction.
• ds_beforeduplication.csv has the data before augmentation.
• augmented_data.csv has the data after augmentation.

book_genre_prediction.ipynb:

This file has the final code containing implementation details of the most accurate models.

Flask:

This folder has the code for web app built using Flask.

Experiments:

This folder has notebook files related to the experiments conducted to evaluate impact of different techniques of classification.

Docs:

This folder has log files of info collected all through the development.

own implementation of models:

This folder has our own implementation of models, created whilst learning. They are not used in implementation.

trial at summary:

This folder has a file where we tried generating summary of a doc. It uses extractive summarization. It is not used in the project.

Links to Drive:

• Pickle Files : All pickle files of trained models

• Multi Label Genre Prediction : Augmented data, code and trained model for Multi Label Genre Prediction.