The Hrant Dink Foundation cordially invites you to the “Detection of Hate Speech on Social Media” contest, which will be held as part of the Computational Social Sciences Session of the Signal Processing and Communication Applications Congress (SIU). The contest is taking place as part of the project"Utilizing Digital Technology for Social Cohesion, Positive Messaging and Peace by Boosting Collaboration, Exchange and Solidarity" sponsored by the European Union and Friedrich Naumann Foundation and in collaboration with Boğaziçi University and Sabancı University.
The goal of the Computational Social Sciences Congress is to raise awareness of hate speech, accelerate the development of interdisciplinary approaches for automated hate speech detection and analysis. Within the framework of the forum, the contest titled “Detection of Hate Speech on Social Media” will be organized, with the aim of cultivating innovative methods for detection of hate speech in Turkish-language texts. At the end of the contest, a joint workshop will be held where the participants will present their methods and promote cooperation and research in the field. The workshop aims to observe the level of improvement in Turkish hate speech detection.
The dataset to be used during the contest includes 5000 tweets in Turkish, containing 3 different topics (Israel-Palestine conflict, opposition towards Greeks and opposition towards refugees/migrants.) The data will be shared with the contestants, and the methods developed by the contestants will be evaluated based on the pre-determined criteria and performance measures. The elected method will be introduced for the benefit of CSOs and online platforms, especially for use by online media and academia.

Process: The training set will be shared with the contestants. Applications can be submitted for one or more of the sub-problems determined within the shared set.

 

 Processing and Communication Applications Congress
5-8 July 2023

Site of the Congress
ITU Ayazaga Campus

Who can participate?

  • Related academics, specialists, students and civil social organization employees.
  • To enroll, you may complete this form until 17 May 2023.
  • For any iSignalnquiries about the contest, you can contact us by email to This email address is being protected from spambots. You need JavaScript enabled to view it..


Agenda for the contest and workshop:

  • Deadline for enrollment: 17 May 2023
  • Deadline for sending out the statement for those willing to participate in the workshop: 31 May 2023
  • Sharing the test set: 1 June 2023
  • Collecting the test results: 2 June 2023
  • Announcing the result of contest: 3 June 2023
  • Announcing the elected statements: 18 June 2023

Official evaluation will be carried out over 4 specific sub-sets.

The method developed by the contestants will be tested over the various data sets below:

  1.  Prediction of hate speech categories through the Israel-Palestine conflict dataset. (Greek opposition set can also be used)
  2. Prediction of the severity of hate speech through the Israel-Palestine conflict dataset. (Greek opposition set can also be used)
  3. Prediction of hate speech category through refugee dataset.
  4. Prediction of the severity of hate speech through the immigrant dataset.

Evaluation Criteria:
The developed methods will be evaluated over the Kaggle platform with independent test sets and the performance of various approaches will be compared with suitable matrixes. (Macro F1 score and average square error.) The contestants are expected to share the code file that generates the csv file they have uploaded to the Kaggle platform, with another form. Contestants are also encouraged to send a statement to be evaluated at the private session. The statements sent out will firstly be evaluated for academic originality. The terms of the contest and reward will be provided in detail before the contest.
The most successful group for each sub-problem of the contest will be awarded €750.

 


This project is financed by the European Union.