Sentiment Analysis - Text Classification with Universal Embeddings
- Category: Natural Language Processing
- Tools Used: Tools and Libraries Used: NumPy, Pandas, Contractions, BeautifulSoup, unicodedata, regular expressions, contractions, beautifulsoup4, TensorFlow, TensorFlow Hub, Neural-Net Language Model (nnlm-en-dim128), Universal Sentence Encoder (USE), TensorFlow Estimators, DNNClassifier, pandas_input_fn, Matplotlib, Seaborn, scikit-learn.
- Project URL: https://github.com/Shubhkirti24/NLP/blob/main/ Transfer_learning_NLP_sentence_encoders.ipynb
Project Summary
This project focuses on performing sentiment analysis using transfer learning with universal sentence embeddings. The goal is to build supervised sentiment analysis models to classify movie reviews from the IMDB Large Movie Review Dataset into positive or negative sentiments. The project leverages pre-trained sentence encoders from TensorFlow Hub, specifically the Neural-Net Language Model (nnlm-en-dim128) and the Universal Sentence Encoder (USE), to enhance the text classification process.