mohanenm/movie-classifier

sentiment classifier in python(and r :) ) @loyolachicagocode

movie-review-classifier

used NLTK to classify movie reviews as positive or negative(sentiment analysis hw for nlp course).

How to use

1. install Natural langauge Toolkit Mac/Linux, run 'sudo pip install -U nltk'

2. In Terminal clone from GH, 'git clone https://github.com/mohanenm/movie-classifier'

3. run, on the command line, 'python ./classifier.py'

4. wait a bit and get your results

5. Fork and Modify it!

How to mess with it

1. Fork it, make it your own

2. Look at the comments on where to test different reviews

3. email me with any questions

Code source

1. https://www.twilio.com/blog/2017/09/sentiment-analysis-python-messy-data-nltk.html

2. NLTK DOCUMENTATION

You will be working with movie review data, which you can download here:https://github.com/dennybritz/cnn-text-classification-tf/tree/master/data/rt-polaritydata

• find two files there (one with positive and one with negative reviews).
• build a binary classifier that will perform movie review classifica-tion automatically.
• 1.split data randomly into training (70%), development (15%)and test (15%) sets. (10 points)
• 2.Download and install LibSVM from https://www.csie.ntu.edu.tw/~cjlin/libsvm/. Convert the sentiment data into LibSVM sparse format. (30points) find some guidelines by following these links:https://www.csie.ntu.edu.tw/~cjlin/libsvm/faq.html#/Q03:_Data_preparationhttps://stats.stackexchange.com/questions/61328/libsvm-data-format
• 3.Use LibSVM command-line tools such as svm-train and svm-predict totrain and evaluate a linear SVM model on your development set. Tune alinear SVM by trying a few different C values such as 0.0001, 0.001, . . . ,1000, 10000 on thedevelopmentset. Report the accuracy you obtainedfor each model. (30 points)
• 4.Evaluate your best model on the test set. Report the final accuracy ofyour model on thetestset. (30 points)