multi_class_tweet_sentiment_analyzer

Objective

Creat an artificial neural network which reads in tweets categorized as any of several emotions and makes predictions on new tweets. Produces as the final project for UA Neural Networks in Fall 2019.

Methods

Model Architecture

The creation of the project got me familiar with the process of building Neural Networks with the Keras API. For this application I used a Recurrent Neural Network with GRU units as they are quire powerful in Natural Language Processing tasks. The metric we optimized for was Jaccard Similarity. I acheived a 0.52 (52%) Jaccard Similarity score, placing me among the top 20% performers in this class.

The Jaccard Index is given as the intersection of two sets / the union of two sets as follows:

Below is the original project description from the class

Objectives

The learning objectives of this assignment are to:

1. build a neural network for a community task
2. practice tuning model hyper-parameters

Read about the CodaLab Competition

You will be participating in a class-wide competition. The competition website is:

https://competitions.codalab.org/competitions/20980?secret_key=5cba3f22-a7d8-4744-8d07-85b01e604967

You should visit that site and read all the details of the competition, which include the task definition, how your model will be evaluated, the format of submissions to the competition, etc.

Create a CodaLab account

You must create a CodaLab account and join the competition:

1. Visit the competition website.

2. In the upper right corner of the page, you should see a "Sign Up" button. Click that and go through the process to create an account. Please use your @email.arizona.edu account when signing up. Your username will be displayed publicly on a leaderboard showing everyone's scores. If you wish to remain anonymous, please select a username that does not reveal your identity. Your instructor will still be able to match your score with your name via your email address, but your email address will not be visible to other students.

3. Return to the competition website and click the "Participate" tab, where you should be able to request to be added to the competition.

4. Wait for your instructor to manually approve your request. This may take a day or two.

5. You should then be able to return to the "Participate" tab and see a "Submit / View Results" option. That means you are fully registered for the task.

Clone the repository

Clone the repository created by GitHub Classroom to your local machine:

git clone https://github.com/ua-ista-457/graduate-project-<your-username>.git

Note that you do not need to create a separate branch as in previous assignments (though you're welcome to if you so choose). You are now ready to begin working on the assignment.

Write your code

You should design a neural network model to perform the task described on the CodaLab site. You must create and train your neural network in the Keras framework that we have been using in the class. You should train and tune your model using the training and development data that is already included in your GitHub Classroom repository.

You may incorporate extra resources beyond this training data, but only if you provide those same resources to all other students in the class by posting the resource in the #graduate-project channel on the class's Slack workspace: http://ua-ista457-fa19.slack.com

There is some sample code in your repository from which you could start, but you should feel free to delete that code entirely and start from scratch if you prefer.

Test your model predictions on the development set

To test the performance of your model, the only officially supported way is to run your model on the development set (included in your GitHub Classroom checkout), format your model predictions as instructed on the CodaLab site, and upload your model's predictions on the "Participate" tab of the CodaLab site.

Unofficially, you may make use of scikit-learn's jaccard_similarity_score to test your model locally. But you are strongly encouraged to upload your model's development set predictions to the CodaLab site many times to make sure you have all the formatting correct. Otherwise, you risk trying to debug formatting issues on the test set, when the time to submit your model predictions is much more limited.

Submit your model predictions on the test set

When the test phase of the competition begins (consult the CodaLab site for the exact timing), the instructor will release the test data and update the CodaLab site to expect predictions on the test set, rather than predictions on the development set. You should run your model on the test data, and upload your model predictions to the CodaLab site.