Google Bindi: Hateful Signals and where to find them?

Google Bindi Summit 2023

Transcript

1. Hateful Signals
   and where to
   find them?
   -Sarah Masud,
   Phd Student @IIIT-Delhi
   

   View Slide

2. Disclaimer
   Subsequent content may contain extreme language (verbatim from social media), which does not
   reflect the opinions of myself or my collaborators. Reader’s discretion is advised.
   

   View Slide

3. Workflow for Analysing and Mitigating Online Hate Speech
   [1]: Tanmoy and Sarah, Nipping in the bud: detection, diffusion and mitigation of hate speech on social media, ACM SIGWEB Winter, Invited Publication
   Fig 1: The various input signals
   (red), models (green) and user
   groups (blue) involved in
   analysing hate speech. [1]
   

   View Slide

4. Why Context is important for Hate Speech?
   ● No clear definition of hate speech.
   ● Subjective annotation from the point
   of view of NLP modeling.
   ● No standard list of vulnerable groups.
   Endogenous Signals
   Exogenous Signals
   

   View Slide

5. Exogenous Signals: News articles & Topical Affinity
   Fig 1: Hatefulness of different users towards different
   hashtags/topics in RETINA [1] Fig 2: Exogenous attention Model RETINA [1]
   XN: News Headline
   XT: Incoming Tweet
   [1]: Masud et al., Hate is the New Infodemic: A Topic-aware Modeling of Hate Speech Diffusion on Twitter, ICDE 2021
   

   View Slide

6. Fig 1: Motivation for Auxiliary Data Signals in Hate Speech Detection [1]
   Endogenous Signals: User’s interaction on platform
   [1]: Kulkarni et al., Revisiting Hate Speech Benchmarks: From Data Curation to System Deployment, KDD 2023
   

   View Slide

7. More about our work:
   https://sara-02.github.io/publications/
   _themessier on Twitter
   lcs2lab on Twitter
   

   View Slide