Sentiment Predictor

About

Deep learning model trained on IMDB dataset to predict sentence sentiment with around 89% accuracy. The model was also trained on a list of popular positive and negative words becouse many of those are missing from IMDB dataset (duh). Ready for deployment on Heroku with Flask.

Usage

Local setup

• run wsgi.py
• Open postman or any other program for testing API's.
• POST on endpoint: http://127.0.0.1:5000/predict

or just use my deployed model on Heroku

• make POST request on endpoint: https://sentiment-prediction-deepl.herokuapp.com/predict with json that looks like some of those below

Examples

(prediction 0 - negative sentence, 1 - positive sentence)

{
    "text": ["The acting is terrible, plot is boring and predictable. What a waste of time.."]
}

response: [{"prediction":0.0,"probability":0.995847702}]  

 {
    "text": ["Great acting, amazing cast. Movie is not trivial and not for an average viewer. Pushes to think and ponder on the meaning of life."]
}

response: [{"prediction":1.0,"probability":0.9719628692}]


{
    "text": ["The pancakes were out of this world, I've never eaten something so tasty in my life"]
}

response: [{"prediction":1.0,"probability":0.9649505019}]  

{
    "text": ["Staff doesn't care about the customer, had to wait for 30 minutes till somebody showed up. Huge disappointment."]
}

response: [{"prediction":0.0,"probability":0.8519150615}] 

Or you can send multiple text samples

{ 
    "text": ["The acting is terrible, plot is boring and predictable. 
                What a waste of time..", 
                "A very nice movie", "I like drinking beer at the sunset."] 
} 

response: [{"prediction":0.0,"probability":0.995847702},
            {"prediction":1.0,"probability":0.8738321066},
            {"prediction":1.0,"probability":0.9644991755}]

Model