digiturk-sentiment-analysis

A flask application that uses Tensorflow Keras Sequential model to classify Turkish text tweets as positive or negative. You can try out the project at https://digiturk-sentiment-analysis.herokuapp.com/

Here is what the app preview looks like:

As it can be seen in the picture above, the machine learning model has classified the word "iğrenç" (which translates to "disgusting")as negative with 63% probablitiy . Here are some other examples:

Here the phrase "çok güzel olmuş ellerinize sağlık" translates to "This is wonderful good job". As this is a positive sentiment, it is classified as positive.

Here the phrase "bok gibi olmuş bir daha olmasın" translates to "This is so bad like shit, please don't let it happen again". As this is a negative sentiment, it is classified as negative.