Maximilian Maurer
@mmmaurer.bsky.social
PhD student at @gesis.org & @hhu.de, computational linguist, researching linguistic factors in (annotation) disagreement and language model behavior.
Addendum: If you’re interested in assessing text properties, you may want to take a look at 🧝elfen, a python package I’ve written to facilitate such analyses on an extensive set of linguistic features and large numbers of text items
github.com/mmmaurer/elfen
github.com/mmmaurer/elfen
GitHub - mmmaurer/elfen: A python package to efficiently extract linguistic features for text/NLP datasets
A python package to efficiently extract linguistic features for text/NLP datasets - mmmaurer/elfen
github.com
November 3, 2025 at 11:55 PM
Addendum: If you’re interested in assessing text properties, you may want to take a look at 🧝elfen, a python package I’ve written to facilitate such analyses on an extensive set of linguistic features and large numbers of text items
github.com/mmmaurer/elfen
github.com/mmmaurer/elfen
It was a very fun and interesting project with excellent collaborators @esradonmez.bsky.social @gabriellalapesa.bsky.social and Agnieszka Falenska
November 3, 2025 at 11:55 PM
It was a very fun and interesting project with excellent collaborators @esradonmez.bsky.social @gabriellalapesa.bsky.social and Agnieszka Falenska
👤🆚🤖
Can we use those differences to effectively and interpretably detect LLM-generated argumentative content?
Turns out we can! Linguistic and argumentative “fingerprints” of human written and model-generated arguments are linearly separable!
Can we use those differences to effectively and interpretably detect LLM-generated argumentative content?
Turns out we can! Linguistic and argumentative “fingerprints” of human written and model-generated arguments are linearly separable!
November 3, 2025 at 11:55 PM
👤🆚🤖
Can we use those differences to effectively and interpretably detect LLM-generated argumentative content?
Turns out we can! Linguistic and argumentative “fingerprints” of human written and model-generated arguments are linearly separable!
Can we use those differences to effectively and interpretably detect LLM-generated argumentative content?
Turns out we can! Linguistic and argumentative “fingerprints” of human written and model-generated arguments are linearly separable!
📝 LLMs generate more formulaic, textbook-like, less lexically rich, more explicitly respectful counter-arguments than humans.
They also more closely match the linguistic style and argumentative properties of the original posts they are countering!
They also more closely match the linguistic style and argumentative properties of the original posts they are countering!
November 3, 2025 at 11:55 PM
📝 LLMs generate more formulaic, textbook-like, less lexically rich, more explicitly respectful counter-arguments than humans.
They also more closely match the linguistic style and argumentative properties of the original posts they are countering!
They also more closely match the linguistic style and argumentative properties of the original posts they are countering!
We assess linguistic and argumentative differences in counter-arguments between humans and LLMs using
👤 Reddit ChangeMyView comments
🤖 LLM-generated responses to the same original reddit posts
We find marked differences between them, both on a distributional and a parallel level!
👤 Reddit ChangeMyView comments
🤖 LLM-generated responses to the same original reddit posts
We find marked differences between them, both on a distributional and a parallel level!
November 3, 2025 at 11:55 PM
We assess linguistic and argumentative differences in counter-arguments between humans and LLMs using
👤 Reddit ChangeMyView comments
🤖 LLM-generated responses to the same original reddit posts
We find marked differences between them, both on a distributional and a parallel level!
👤 Reddit ChangeMyView comments
🤖 LLM-generated responses to the same original reddit posts
We find marked differences between them, both on a distributional and a parallel level!
Reposted by Maximilian Maurer
*Preprocessing and Analyzing Web Tracking Data* by Frank Mangold, Helena Rauxloh & @sebstier.bsky.social
🗓️ 04-05 May 2026
🏢 Hybrid: Cologne | Online
🌐 t1p.de/WebTrackingD...
🗓️ 04-05 May 2026
🏢 Hybrid: Cologne | Online
🌐 t1p.de/WebTrackingD...
November 3, 2025 at 1:28 PM
*Preprocessing and Analyzing Web Tracking Data* by Frank Mangold, Helena Rauxloh & @sebstier.bsky.social
🗓️ 04-05 May 2026
🏢 Hybrid: Cologne | Online
🌐 t1p.de/WebTrackingD...
🗓️ 04-05 May 2026
🏢 Hybrid: Cologne | Online
🌐 t1p.de/WebTrackingD...
Reposted by Maximilian Maurer
Ach und er sagt auch noch, dass Shitstorm ein modernes Wort für Pogrome ist. Genau! Und Jana aus Kassel ist Sophie Scholl wegen Kritik. Es ist EXAKT DASSELBE DENKEN wie bei den Schwurblern, die sich zur Pandemie einen Judenstern angeheftet haben. Mach mal Fenster auf Kipp, Richard.
October 17, 2025 at 6:41 AM
Ach und er sagt auch noch, dass Shitstorm ein modernes Wort für Pogrome ist. Genau! Und Jana aus Kassel ist Sophie Scholl wegen Kritik. Es ist EXAKT DASSELBE DENKEN wie bei den Schwurblern, die sich zur Pandemie einen Judenstern angeheftet haben. Mach mal Fenster auf Kipp, Richard.