"D-E", pursuing PhD @oiioxford | researching platform governance, content moderation, human-computer interaction | https://teapaw.org/t.co/CKXYWkF0jT
Joined April 2019
- Tweets121
- Following894
- Followers437
- Likes973
Pinned Tweet
Over time we see recurring themes of securitization and public morality as both the platform and local states, despite not being well-positioned, push the lines of moderation. - The implications for users' digital rights are profound. doi.org/10.1002/poi3.388
Diyi Liu retweeted
New blog! @oiioxford doctoral researchers @deeliu97
@ManuelTonneau and Juliette Zaccour propose a series of recommendations for effective data access and data governance in light of the EU’s Digital Service Act.
Read the full article:
oii.ox.ac.uk/news-events/oii…
Another impressive study on multilingual/cultural hate speech detection by @ManuelTonneau, this time based on a complete 24h Twitter dataset. What does it imply for balancing the costs and necessity of human-in-the-loop HS moderation? Do check out the paper & the HateDay dataset:
Can we detect #hatespeech at scale on social media?
To answer this, we introduce 🤬HateDay🗓️, a global hate speech dataset representative of a day on Twitter.
The answer: not really! Detection perf is low and overestimated by traditional eval methods
arxiv.org/abs/2411.15462
🧵
4
528
If you're working on cross-cultural hate speech detection or related research, do check out the supersets collection by @ManuelTonneau!!
Our #hatespeech supersets, combining all 🤬 corpora in 8 lang, reached 2K downloads on 🤗🎉
To enable more cross-cultural 🤬 research, we release:
- author country for English, Arabic and Spanish posts from our @WOAHWorkshop 📜
- 3 country sets 🇮🇳🇳🇬🇰🇪
huggingface.co/collections/m…
1
268
BIG publication!!!
Big publication alert!
Happy to announce my first first-author paper being out in the wild: doi.org/10.12758/mda.2024.08
Written together with @dscheykopp & @theoaraujo we 1.) propose the methodological concept of augmented Data Download Packages (aDDPs), ...
2
205
Diyi Liu retweeted
How ByteDance balances competing priorities around protecting children. The analysis reveals that TikTok employs three interconnected strategies: scaffolding rules by age and risk, segmenting content with tailored policies, and siloing features from children.
Out this week in @IJCS_journal -- how do TikTok's community guidelines protect children, reflect the company's latent values, and advance its strategic priorities? Thanks to @jinlee_media + @wishcrys + @tamaleaver for this SI!
Key takeaways in the 🧵:
doi.org/10.1177/136787792412…
Happened to me as well. These articles are only discoverable through Google Scholar page, not on the original publication website. And the indexing system makes it nearly impossible to curate citations (by myself
This Post is from an account that no longer exists. Learn more
4
417
Diyi Liu retweeted
Next up from the latest in #HMC...this HMC community paper explores authenticity.
Thanks to @Katrin_ee, @DrJihyunKim, @margotvdgoot, and Andrew Prahl for leading this massive effort. Be sure to check it out.
I am so proud to share our first HMC community paper in @HMCJournal, where 22 scholars collaborated on "What HMC Teaches Us About #Authenticity": stars.library.ucf.edu/hmc/vo…
Thank you to everyone who contributed to this incredible journey! 💫
#HMCommunity #ProudMoment #ThankYou
Accounting for cultural contexts in hate speech moderation is an ambitious yet crucial learning process that we must continually reflect on. Building on previous efforts in the field, my brilliant colleague @ManuelTonneau has led important analysis on HS detection resources👇
What is deemed #hatespeech can vary across cultures, but 🤬 datasets are often built only at the language level 🗣️, masking potential cultural biases. Our new @WOAHWorkshop paper reveals large cultural representation gaps in 🤬 datasets using 🗣️ and 🌍 as cultural proxies.
🧵
1
2
520
It was indeed a great learning experience for me too! Please do check out the full paper here: arxiv.org/abs/2404.17874
99
Diyi Liu retweeted
What has TikTok done to implement Project Texas since it first publicized the plan? Earlier this week, after TikTok challenged the constitutionality of the “divest or ban” law, I received a briefing on progress, and wrote up what I learned for @lawfare.
lawfaremedia.org/article/has…
Diyi Liu retweeted
/1 Today, we're making it easier to tell when content is AI-generated, by being the first video-sharing platform to start implementing @C2PA_org Content Credentials and labeling AIGC from other platforms. 🧵
Diyi Liu retweeted
Bytedance is exploring scenarios for selling #TikTok without algorithms
Hell of a scoop by @erinkwoo @QianerLiu & @JuroOsawa
Subscribe to @theinformation to stay ahead of curve
theinformation.com/articles/…
🈹️🈲Once again, TikTok's attempt to present itself as an entertaining playground turns into a political battleground. My latest @policyinternet explores how TT’s controversies unfold in South and Southeast Asia, where huge market potential intersects with delicate speech norms.
Kicking off April with a change of scenery and an electrifying new adventure. From the 7th, I’m honored to begin an Open Research Fellowship at @JWI_Berlin. Look forward to connecting with many friends and time for my indulgence in 🦋🍺🥨🎼👯♀️and 📝(?)
3
25
1,190
Diyi Liu retweeted
Excited to run for #IAMCR Vice President! Full Professor @NUSingapore, I've been with IAMCR since 2011, serving as Vice-Chair & Co-Chair of #iamcrCPT. My vision: boost #Inclusivity, global #impact, & voice to action. Watch my video: [shorturl.at/hzIU0] vote at #iamcr2024!
