Sentiment Polarity CoreML Demo

A Demo application using CoreML framework for sentiment polarity analysis.

Model

CoreML model was converted from Scikit-learn Pipeline using coremltools python package.

The model is based on LinearSVC classifier and is able to distinguish between positive and negative sentences with best CV score = 0.801013024602. It was trained using Epinions.com dataset with reviews of products and services. Accuracy can be improved by using TfidfVectorizer for feature extraction, but it's not supported by coremltools at the moment.

Requirements

• Xcode 9
• iOS 11

Installation

git clone https://github.com/cocoa-ai/SentimentCoreMLDemo.git
cd SentimentCoreMLDemo
open SentimentPolarity.xcodeproj/

Build the project and run it on a simulator or a device with iOS 11.

Conversion

cd Convert
python sentiment.py

Author

Vadym Markov, markov.vadym@gmail.com

Credits

• Dataset with Epinions.com reviews was taken from HW3: Sentiment Analysis

References

• scikit-learn
• Apple Machine Learning
• CoreML Framework
• coremltools