flowchart LR subgraph ways[acting] direction LR speech writing behavior end subgraph hurt[that aims to hurt] direction LR attacks lang[pejorative language] end subgraph targets[people] direction LR individuals groups end subgraph who[based on who they are] direction LR race gender etc. end ways --> hurt --> targets --> who
The Social Media Hate Speech Barometer
Making-of
Sebastian Sauer, Alexander Piazza, Sigurd Schacht
University of Ansbach
July 14, 2023
Project objective
To provide applied researcher a template for speech classification
Definition of hate speech1
Hate speech as a menace to society
flowchart LR sm[Social Media] --> hs[hate speech] hs --> dem[destabilizes democracy] hs --> civ[civil rights] hs --> mh[mental health] hs --> ph[psychosocial health]
Calvert (1997), Cinelli et al. (2021), Castaño-Pulgarín et al. (2021), Chan, Ghose, and Seamans (2016)
Machine learning for detecting hate speech? See Hartvigsen et al. (2022), Velankar, Patil, and Joshi (2022)
Overview on design ideas
Project management via targets
Machine learning via tidymodels in
Collaboration platform via Github
1. targets manages outdated computational objects
2. Machine learning standard pipeline
flowchart LR
prepData[Import and </br>prepare data] --> folds
subgraph folds[for each fold do]
subgraph tune[for each tune value do]
direction TB
prep2[prepare data] --> fit[fit model] --> predict[predict on</br>assessment data]
end
end
folds --> fit2[fit on train data]
fit2 -->pred2[predict on test data]
pred2 --> performance[assess</br>performance]
2. library(tidymodels)
d_split <- initial_split(mtcars)
ranger_recipe <- recipe(formula = am ~ ., data = mtcars)
ranger_spec <-
rand_forest(mtry = tune(), min_n = tune(), trees = 1000) %>%
set_mode("classification") %>%
set_engine("ranger")
rsmpl <- vfold_cv(d_train)
ranger_workflow <- workflow() %>%
add_recipe(ranger_recipe) %>%
add_model(ranger_spec)
ranger_tune <- tune_grid(ranger_workflow,
resamples = stop("add your rsample object"),
grid = stop("add number of candidate points"))2. Machine learning pipeline
3. Collaborate via git(hub)
That’s all folks! 👋
QR-code with link to the slides
Sebastian Sauer
sebastian.sauer(å†)hs-ansbach.de
https://sebastiansauer.github.io/hate-speech-barometer/talks/bmt-2023
Technical details
Date of last update: 2023-07-14 08:01:26.
setting value
version R version 4.2.1 (2022-06-23)
os macOS Big Sur ... 10.16
system x86_64, darwin17.0
ui X11
language (EN)
collate en_US.UTF-8
ctype en_US.UTF-8
tz Europe/Madrid
date 2023-07-14
pandoc 3.1.1 @ /Applications/RStudio.app/Contents/Resources/app/quarto/bin/tools/ (via rmarkdown)References
Calvert, C. 1997. “Hate Speech and Its Harms: A Communication Theory Perspective.” Journal of Communication 47 (1): 4–19. https://doi.org/10.1111/j.1460-2466.1997.tb02690.x.
Castaño-Pulgarín, Sergio Andrés, Natalia Suárez-Betancur, Luz Magnolia Tilano Vega, and Harvey Mauricio Herrera López. 2021. “Internet, Social Media and Online Hate Speech. Systematic Review.” Aggression and Violent Behavior 58 (May): 101608. https://doi.org/10.1016/j.avb.2021.101608.
Chan, Jason, Anindya Ghose, and Robert Seamans. 2016. “The Internet and Racial Hate Crime: Offline Spillovers from Online Access.” Management Information Systems Quarterly 40 (2): 381–403. https://aisel.aisnet.org/misq/vol40/iss2/8.
Cinelli, Matteo, Andraž Pelicon, Igor Mozetič, Walter Quattrociocchi, Petra Kralj Novak, and Fabiana Zollo. 2021. “Dynamics of Online Hate and Misinformation.” Scientific Reports 11 (1, 1): 22083. https://doi.org/10.1038/s41598-021-01487-w.
Hartvigsen, Thomas, Saadia Gabriel, Hamid Palangi, Maarten Sap, Dipankar Ray, and Ece Kamar. 2022. “ToxiGen: A Large-Scale Machine-Generated Dataset for Adversarial and Implicit Hate Speech Detection.” July 14, 2022. https://doi.org/10.48550/arXiv.2203.09509.
Velankar, Abhishek, Hrushikesh Patil, and Raviraj Joshi. 2022. “A Review of Challenges in Machine Learning Based Automated Hate Speech Detection.” September 12, 2022. https://doi.org/10.48550/arXiv.2209.05294.