DETECTING BOOK GENRES FROM THE BOOK DESCRIPTION USING A FNN

Keywords

Multi class text classification, book genre detection, machine learning, FNN, BlurbGenreCollection-EN

Dataset:

92.000 (english) Books from the BlurbGenreCollection-EN (https://www.inf.uni-hamburg.de/en/inst/ab/lt/resources/data/blurb-genre-collection.html)

5 Attributes:

• Title

• Author

• ISBN

• Genre

• Description

Aim:

Build a classifier that predicts the book‘s genre

Approach:

1. Explore Dataset & Preprocessing
2. Split Data in Train, Test & Val Set
3. Calculation of Baseline Results for Comparison
4. Text Cleaning
5. Vectorization (TF-IDF)
6. Label Encoding
7. Hyperparameter Tuning
8. Build two-layer-Perceptron