PCD (Polish Cyberbullying Dataset)

The Polish Cyberbullying Dataset is a valuable resource for studying harmful online phenomena, specifically cyberbullying and hate speech in the Polish language. Let's delve into the details:

1. Dataset Overview:

   • The dataset contains tweets that have been annotated with labels indicating whether they fall into the category of cyberbullying/harmful or non-cyberbullying/non-harmful content.
   • These tweets were collected from publicly available Twitter discussions.
   • Minimal preprocessing has been applied to the tweets, primarily focusing on cases where information about a private individual is revealed to the public.

2. Purpose and Significance:

   • The lack of research on cyberbullying detection in the Polish language prompted the creation of this dataset.
   • Researchers aim to automatically detect harmful content on the internet and report it to service providers for further analysis and removal.
   • The dataset facilitates the development and evaluation of classification methods for identifying cyberbullying-related narratives.

3. Tasks Available:

   • Users can experiment with their classification algorithms to determine whether an internet entry (tweet) is part of a cyberbullying narrative or not.
   • The primary goal is to classify tweets into either cyberbullying/harmful or non-cyberbullying/non-harmful categories, achieving high precision, recall, balanced F-score, and accuracy.
   • An additional sub-task focuses on differentiating between various types of harmful information, such as cyberbullying or hate speech.

4. Example Format of Tweets:

   • Each tweet is represented as follows:

     "contents", class (1=harmful, 0=non-harmful, etc.)
     

5. Accessing the Dataset:

   • The dataset is available on GitHub under the repository named **"cyberbullying-Polish"**¹.
   • It includes annotated labels, scoring scripts, and raw tweets for research purposes.

This dataset contributes significantly to addressing the growing problem of harmful online behavior, providing researchers with a foundation for developing automatic cyberbullying detection methods in the Polish language ².

Source: Conversation with Bing, 3/16/2024 (1) GitHub - ptaszynski/cyberbullying-Polish: This dataset contains tweets .... https://github.com/ptaszynski/cyberbullying-Polish. (2) Expert-Annotated Dataset to Study Cyberbullying in Polish Language - MDPI. https://www.mdpi.com/2306-5729/9/1/1. (3) Data | Free Full-Text | Expert-Annotated Dataset to Study Cyberbullying .... https://www.mdpi.com/2306-5729/9/1/1/notes. (4) undefined. http://poleval.pl/tasks/task6.