Sentiment Analysis of Persian Comments

Implementation of NLP models for classification of the Persian comments' sentiment in Python.

Requirements

see requirements.txt and use pip install -r requirements.txt for installation.

1. numpy for some calculations
2. pandas for data read/write
3. scikit-learn for classfiers
4. gensim for Word2Vec model
   
   

About Sentiment Analysis

Sentiment analysis (also known as opinion mining or emotion AI) is the use of natural language processing, text analysis, computational linguistics, and biometrics to systematically identify, extract, quantify, and study affective states and subjective information. Sentiment analysis is widely applied to voice of the customer materials such as reviews and survey responses, online and social media, and healthcare materials for applications that range from marketing to customer service to clinical medicine [Wiki]. Following figure shows the workflow of sentiment analysis [monkeylearn].

Project structure

• 📂 directory [sentiment_data]

  This directory contains data used in the project. train.csv has labeled comments for training (we split it into train-validation to find the best model). test.csv has unlabeled comments that we should predict the label of each row using our propused model and save the result in test-labeled.csv.

• 📂 directory [nlp_files]

  This directory contains some files for being used in NLP models (e.g. stop-words).

• 📄 [phase1.ipynb]

  Contains steps for finding the best method for being applied to data. In this file, we only use train.csv and split it into train-validation sets. Here is the final result of all 18 models employed in this phase:

Classifiers	TF-IDF	Word2Vec
KNN(n=4)	0.4619	0.6131
KNN(n=8)	0.6940	0.6298
KNN(n=16)	0.7524	0.6238
SVM(linear)	0.7905	0.4774
SVM(poly)	0.6417	0.6452
SVM(rbf)	0.7905	0.6810
XGB(n=50)	0.7214	0.6571
XGB(n=100)	0.7298	0.6786
XGB(n=150)	0.7381	0.6714

• 📄 [phase2.ipynb]

  This file is the implementation of the proposed method found in phase1. The result of applying the proposed model on test.csv can be seen in test-labeled.csv.