README.md

Project is intended to train and test sentiments on mobile phone.

First phase is training the data on device (mobile) and later reading the trained model file.

MODEL_PATH ( internal storage ) :

private static String MODEL_PATH = Environment.getExternalStorageDirectory().toString()+"/SENTIMENT_DATA/";

Trained model File :

private static String ModelFile = "Sentiment.text";

Training data path :

private static String TRAINING_DATA_PATH = Environment.getExternalStorageDirectory().toString()+"/SENTIMENT_DATA/training/";

training/POS/pos.txt (create POS folder and keep pos.txt at SENTIMENT_DATA)

training/NEG/neg.txt (create NEG folder and keep neg.txt at SENTIMENT_DATA)

Testing data path :

private static String TESTING_DATA_PATH = Environment.getExternalStorageDirectory().toString()+"/SENTIMENT_DATA/testing/";

testing/POS/pos.txt

testing/NEG/neg.txt

checkUnlabeledData() API helps in cross validation for /testing/ path

Testing API : checkUnlabeledData(String rawText)

Use high end device : Samsung Galaxy S7 and above

OnDevice WordVector : https://github.com/vijay033/Deeplearning4j

Other use cases :

(a) Message Classification on mobile phone (b) Quotes Classification (d) Document Types classification

https://github.com/vijay033/SentimentAnalysis_Android