ISCA's 2025 DML Walkthrough: How to Create a Discourse Dataset from X/Twitter (Challenge #1)

Описание: This tutorial offers a complete walkthrough of Challenge #1 of the 2025 Datathon and Machine Learning Competition on Antisemitism, hosted by Indiana University’s Institute for the Study of Contemporary Antisemitism (ISCA). It guides participants through the key steps for collecting, processing, and annotating social media data from X (formerly Twitter) as part of a hands-on hate speech detection task.

The recording covers all technical and methodological components, including how to search and scrape posts using Bright Data, pre-process the data in Google Colab, and structure your annotations using the Annotation Portal.

All necessary tools, links, and templates are provided below to support an independent and structured workflow.

Chapters:
01:00 Agenda & Objectives

01:47 Prerequisites & Setup

02:26 What is the Datathon?

02:57 Instructions for Challenge #1

03:53 Working With X's Advanced Search Function

08:34 Working With Bright Data

11:28 Working With Google Colab

17:28 Annotation Portal Walkthrough

23:02 Setting Up An Annotation Scheme

29:02 Download Annotated Dataset


Useful Resources and Links

Annotation Portal: https://annotate.osome.iu.edu

Challenge Description (PDF): https://github.com/damieh1/datathon_2...

GitHub ReadMe (tools & setup): https://github.com/AnnotationPortal/D...

Bright Data Scraper (no code): https://brightdata.com/cp/scrapers/no...

Google Colab (preprocessing script): https://colab.research.google.com/git...

Download the full slide deck: https://github.com/damieh1/datathon_2...