A project by

and partners

Centre.My Logo
A project by The Centre and partners


1. What constitutes hate speech?

There are many definitions of hate speech/hateful speech. #TrackerBenci defines hate speech as speech that is public, speech deemed as insulting or threatening, and speech that targets a group trait such as ethnicity, nationality, gender, or sexual orientation.

2. Why track hate speech at all?

Online hate speech and hateful content has become a major concern in societies around the world. There is a growing trend where audiences convene around such content in virtual spaces, and develop or strengthen biases against others. The risks of online hate speech leading to violence has been well researched, but there are also broader risks to inter-group trust and cohesion.

This issue requires a more effective response, backed by a localised and idiomatic understanding of what constitutes hate speech. At The Centre, we believe that responding to hate speech in Malaysia requires detecting it consistently, which can then be used for further work to determine its seriousness, impact, and the appropriate response to it.

3. Does tracking hate speech undermine free speech?

Tracking hate speech does not undermine free speech if it is accompanied by an appropriate and proportionate set of responses. We regard free speech as a fundamental right for all, but its impact on a diverse multicultural society such as Malaysia’s needs to be researched and understood. Tracking hateful speech, and developing better methods for tracking it, is an important part of this endeavour.

4. How does #TrackerBenci work?

The #TrackerBenci software was first ‘trained’ with a list of words commonly used in insulting, threatening or otherwise hateful messages and postings targeting various groups in Malaysia. For this Version 1.0 of #TrackerBenci, the reference list of words comprised English, Bahasa Malaysia and Malaysian colloquial words, selected by The Centre’s diverse (though small) panel of locally-informed researchers who sifted through thousands of tweets to develop the training set.

For V1.0 of #TrackerBenci, the software is trained to differentiate between negative vs. not negative or neutral speech. It is trained further to differentiate between speech deemed insulting vs. speech that is threatening. The software is also trained to identify the targeted group as well as the potential themes underlying the speech, for example whether it perpetuates negative stereotypes or is religion-related or is a violent threat.

5. How can you attribute #TrackerBenci in your work?

You can reference the name “#TrackerBenci” as well as the URL “https://centre.my/trackerbenci” in your work.

6. How can you help improve #TrackerBenci?

Submit your input to the Benci Calculator here. If you’d like to be more involved in the architecture or design of the Calculator/Tracker, do get in touch with us at info@centre.my.

7. Does #TrackerBenci store your data?

The Tracker/Calculator will store inputs of words or phrases for future training and improvement of the underlying software. The Tracker/Calculator does not have access to, nor does it store, any personal data.

#TrackerBenci V1.0 is a prototype by The Centre designed to track Malaysian-context hateful speech with AI. We, and the machine, are still learning - stay tuned!

Developed in partnership with Tellnsearch.