Sentiment Analisys with Tensorflow

This is a sample of sentiment analisys for romanian and english language built up on tensorflow and tflearn

Requirements

• tensorflow 1.8
• tflearn 0.3.2
• numpy 1.14.3
• python 3.6.4

Accuracy

After 10 epochs for romanian dataset :

Training Step: 5360  | total loss: 0.05475 | time: 51.239s
| Adam | epoch: 010 | loss: 0.05475 - acc: 0.9881 | val_loss: 0.53884 - val_acc: 0.8536 -- iter: 17144/17144

After 10 epochs for english dataset :

Training Step: 13620  | total loss: 0.05636 | time: 142.459s
| Adam | epoch: 010 | loss: 0.05636 - acc: 0.9940 | val_loss: 0.18359 - val_acc: 0.9396 -- iter: 43561/43561

How to use it

Predict

Pass en for English or ro for Romanian as arg to command line followed by text

$ python predict.py en "Food is awesome"
negative=0.022818154
positive=0.97718185

$ python predict.py ro "Mancarea este proasta"
negative=0.9629853
positive=0.037014768

Train

Pass en for English or ro for Romanian as arg to command line

$ python train.py en

Datasets

Positive dataset is more numerous than the negative one. This may cause a drop in accuracy.

English dataset

Dataset of reviews from yelp and imdb reviews

Train:

• neg: 10090
• pos: 33532
• total: 43561

Test

• neg: 10228
• pos: 33394
• total: 43622

Romanian dataset

Dataset of products and movies reviews

Train:

• neg: 6370
• pos: 10774
• total: 17144

Test

• neg: 4427
• pos: 5840
• total: 10267