Skip to content

Latest commit

 

History

History
69 lines (42 loc) · 1.59 KB

README.md

File metadata and controls

69 lines (42 loc) · 1.59 KB
Raw

Popular Quotes Author Classifier Using NLP

Goal

The main goal of the project is to use NLP techniques to predict what is the name of an author from his quotes.

Dataset

Link : https://www.kaggle.com/datasets/rafsunahmad/popular-quotes-author-classifier

Description

This project consists of code that uses NLP techniques for author name prediction from the specific author's quotes. The dataset consists of two columns, one for Quotes and the other for the name of the author.

What I Had Done

  1. Extract the data from the csv file using pandas
  2. Apply the preprocessing steps that include missing values removal, duplicate removal, data augmentation and a few other minor steps
  3. Use the DL Models to make the predictions

Models Used

  1. BERT
  2. Word2Vec + RNN(LSTM)
  3. TF-IDF + Softmax Regression

Libraries Needed

  1. tensorflow
  2. pandas
  3. nlpaug
  4. numpy
  5. transformers
  6. scikit-learn
  7. gensim
  8. matplotlib
  9. nltk

Visualization

All the authors and their corresponding number of quotes in the dataset

Image

Length Distribution of Words

Image

Pie Chart of top 20 Words

Image

Accuracies

  1. BERT - 91.32%
  2. Word2Vec + RNN(LSTM) - 50.00%
  3. TF-IDF + Softmax Regression - 86.57%

Conclusion

From the above accuracies, we conclude that BERT is the model with the best performance.

Your Name

Hi, I am Iman Kalyan Chakraborty, a passionate ML and DL developer from Kolkata. Here are my socials :

Twitter / X - https://twitter.com/ikc1975

LinkedIn - https://www.linkedin.com/in/imankalyanchakraborty/