Turkish Hate Speech Detection Model
This model is prepared by Turkish text data to identify hate speech towards a specific ethnicity, nationality, religion, sexuality, disability, or class and prevent this kind of malicious intent in the media.
As the interest in social media grows and the shares of users increase, the number of malicious discourses and hateful posts increases with the same rate. By the "Turkish Hate Speech" model prepared for content creators who want to prevent such intent on social media channels, Turkish text data can be categorized as "hate speech" and "not hate speech" quickly and easily. It does not matter how many rows the data has. After testing this model, the content creators can review their API documentation and integrate the model into their media.
About Training Set
The prepared dataset for the Turkish Hate Speech Detection Model was collected from Kimola's Analytics Platform and trained over a freely available dataset consists of 2,092 social media contents. Data was labeled as "hate speech" and "not hate speech." Before the last version of this dataset was reached, the training dataset was tested with four different data consisting of random social media contents. Each wrongly labeled row was exemplified until correct labeling was achieved.