2023 Datathon Competition

Announcing the 2023 Datathon & Machine Learning Competition Winners

We are thrilled to announce that we received excellent submissions for the 2023 Datathon & Machine Learning Competition and are proud to announce the winners. Congratulations!

Team 1 (Liv Sandberg, Meherprasad Bade, Shruti Sharma, Aditya Agarwal, and Xena Doris Pereira) stood out for its sophisticated models in the hackathon part, followed by Team 6 (Juli Thach, Swapnali Gujar, Trenton Reeves, and Yang Zhou). The order was reversed in the datathon part. Team 6 annotated by far the most tweets correctly, followed by Team 1.

Team 1 and Team 6 both earned the first prize.

Team 4 (Asher William McCready, Jesse Levitas, Mayeesha Farjana, Natanya Bath, Ranvir Singh Virk, and Himanshu Gupta) did well in both parts and received the third prize.

The evaluation of the results was presented at the opening of Indiana University's Luddy Artificial Intelligence Center on March 8, 2023.

We thank all teams for their participation.

Visit our GitHub Page

Winning teams’ hackathon solutions:

Team 1 (Liv Sandberg, Meherprasad Bade, Shruti Sharma, Aditya Agarwal, and Xena Doris Pereira)

Team 1 used nltk, pickle, wordnet, and Scikit-learn's for training. They built models for both datasets (Gold Standard and Class Data) and merged them. The models are for both biased messages and calling-out and an additional model is for biased messages only. They also built two models to clean the data. They used 80% of the data to train the model and 20% to test it. Their scores are close despite the noise in the datatsets. We ran their model on a cleaner dataset which resulted in an excellent F1 score of 0.92. The models are written in python and provided here.

Team 6 (Juli Thach, Swapnali Gujar, Trenton Reeves, and Yang Zhou)

Team 6 used different libraries to accomplish the requirements, including nltk, scikit-learn's, pickle, and Ginger. They have train and test models. In the train models they have a code for cleaning the text and then predicts whether a given text is biased or not. They built another model to predict whether a text belongs to a particular label or not. The text is predicted as belonging to the label or not by using the pre-trained models. The code is well-organized and follows a standard pipeline for sentiment analysis.

Team 4 (Asher William McCready, Jesse Levitas, Mayeesha Farjana, Natanya Bath, Ranvir Singh Virk, and Himanshu Gupta)

Team 4 used different libraries to accomplish the requirements, including sckit-learn’s and Numpy. The code performs data cleaning, preprocessing, and feature extraction on them. The text data is cleaned by removing unwanted characters, converting contractions to their expanded form and removing special characters. The code is well-written, and the approach taken for data cleaning, preprocessing, and feature extraction is standard.

2023 Datathon and Machine Learning Competition

Bias Against Asians, Black People, Jews, Hispanics/Latinx, and Muslims on Social Media: Datathon Competition for High School and Undergraduate Students

Want to learn about online hate speech and how to detect and combat it using Machine Learning?

In these workshops, led by IU professors, participants will learn how to recognize, monitor, and track biased messages on social media (specifically Twitter). Participants will be placed into teams and asked to work together on their own scripts and annotations. The three workshops will be:

1: What is hate speech online? What are manifestations of bias against Asians, black people, Hispanics, Jews, and Muslims?  What is Machine Learning? Why do we need Machine Learning to observe and combat hate speech?  

2: Why do we need manually labeled datasets? How to annotate Tweets? 

3: How to use Machine Learning to work with an annotated corpus?

Participate in:

a series of competitive based workshops coached by Indiana University Professors which include:
- a DATATHON where you learn how to create, label, and analyze corpora 
-  a HACKATHON where you learn how to create and use Machine Learning algorithms

Prizes:

Top three teams will be awarded a $500, $300, or $200 Visa gift card
Each participant will receive a t-shirt for completing the program

Requirements:

No prior knowledge is required, but minimal experience with programming will be useful.
All high school and undergraduate students from all countries are welcome to apply.

Because of the nature of this program, you will need access to a computer and an internet connection. All workshops will be conducted over Zoom.

Dates:

Jan. 15,  2023 @ 1-3 pm ET

Jan. 22,  2023 @ 1-3 pm ET

Jan. 29,  2023 @ 1-3 pm ET

How to get involved:

Visit our GitHub Page

Download Shareable Flyer

Contact:

Professor Jikeli: gjikeli@indiana.edu

Jenna Solomon: jpsolomo@iu.edu

Grace Bland: gcbland@iu.edu