/deep-learning-sentiment-analysis

Sentiment Analysis with gensim, Stanford CoreNLP, and TensorFlow

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

deep-learning-sentiment-analysis

deep-learning-sentiment-analysis is an NLP project that compares three different models for binary sentiment classification.

Data

deep-learning-sentiment-analysis uses Stanford's [Large Movie Review Dataset] (http://ai.stanford.edu/~amaas/data/sentiment/). It consists of sets for positive train, negative train, positive test, and negative test, each of which has 12,500 reviews, along with 50,000 unlabeled reviews for unsupervised learning, for 100,000 total reviews. Each review is comprised of multiple sentences.

Models

deep-learning-sentiment-analysis utilizes three different models for sentiment analysis:

Academic Background

Software Dependencies

deep-learning-sentiment-analysis is written in Python 2.7 in a Jupyter notebook and uses several common software libraries, most notably Stanford CoreNLP, gensim, and TensorFlow. In order to run it, you must install the follow dependencies:

License

This project uses the [Apache 2.0 License] (https://github.com/charlescc9/deep-learning-sentiment-analysis/blob/master/LICENSE).