#TextAsData/#NLP
LightNews LightNews
chrauh.bsky.social
🚨 #Data alert 📣 We just released #ParlLawSpeech – full texts of more than 40k bills, 28k laws, and 3 mio. parliamentary speeches from 7 countries (AT, CZ, DE, DK, ES, HR, HU) and the EU! If you study democracy with #TextAsData / #NLP methods, this is for you! A short 🧵 (1/3) #PoliSkyData #polisky
February 13, 2025 at 5:29 PM
clint0475.bsky.social
📡The next term of the #TaDa Speaker Series starts next week!! We have six amazing speakers on all things #TextAsData/#NLP lined up for you

👇
October 11, 2023 at 8:49 AM
chrauh.bsky.social
🤔 The #EU 🇪🇺 aspires to be a global actor 🌍 — but do other states recognize it as such?

My new study in @intlinteractions.bsky.social develops targeted #NLP / #TextAsData tools to analyze 50 years of foreign policy discourse in the annual #UN General Debate (1970–2020).

A thread (1/n)
July 16, 2025 at 7:57 AM
jbgruber.bsky.social
If you feel uneasy using LLMs for data annotation, you are right (if not, you should). It offers new chances for research that is difficult with traditional #NLP/#textasdata methods, but the risk of false conclusions is high!

Experiment + *evidence-based* mitigation strategies in this preprint 👇
joachimbaumann.bsky.social Joachim Baumann @joachimbaumann.bsky.social · Sep 12
🚨 New paper alert 🚨 Using LLMs as data annotators, you can produce any scientific result you want. We call this **LLM Hacking**.

Paper: arxiv.org/pdf/2509.08825
We present our new preprint titled "Large Language Model Hacking: Quantifying the Hidden Risks of Using LLMs for Text Annotation". We quantify LLM hacking risk through systematic replication of 37 diverse computational social science annotation tasks. For these tasks, we use a combined set of 2,361 realistic hypotheses that researchers might test using these annotations. Then, we collect 13 million LLM annotations across plausible LLM configurations. These annotations feed into 1.4 million regressions testing the hypotheses. For a hypothesis with no true effect (ground truth $p > 0.05$), different LLM configurations yield conflicting conclusions. Checkmarks indicate correct statistical conclusions matching ground truth; crosses indicate LLM hacking -- incorrect conclusions due to annotation errors. Across all experiments, LLM hacking occurs in 31-50\% of cases even with highly capable models. Since minor configuration changes can flip scientific conclusions, from correct to incorrect, LLM hacking can be exploited to present anything as statistically significant.
September 15, 2025 at 1:05 PM
jbgruber.bsky.social
If you do research with/about LLMs, image or audio models, or classic NLP methods, you should check out this call 👇

#textasdata #NLP #llm #commsky #polsky
comptext.bsky.social The COMPTEXT Association @comptext.bsky.social · Jan 12
Submission for #COMPTEXT2026 is still open 𝐮𝐧𝐭𝐢𝐥 𝐅𝐫𝐢𝐝𝐚𝐲.
We are looking forward to receiving your proposals for papers and panels.
We also appreciate it if you can spread the word and circulate shorturl.at/gRg0p!
See you in Birmingham, 23–25 April 2026!
January 12, 2026 at 8:54 AM
chrisguarnold.bsky.social
Stumbled over this pretty cool living bibliography on causal textanalysis. Might be useful to some. #textasdata #causal #nlp #sciencerocks github.com/causaltext/c...
GitHub - causaltext/causal-text-papers: Curated research at the intersection of causal inference and natural language processing.
Curated research at the intersection of causal inference and natural language processing. - GitHub - causaltext/causal-text-papers: Curated research at the intersection of causal inference and natu...
github.com
February 9, 2024 at 11:26 AM
abitter.bsky.social
"Potenziale von #KI für die Personalisierung von (Weiter-)Bildung" - der Vortrag von @afischer1985.bsky.social
ist nun im @zpid.bsky.social Mediacenter frei verfügbar!

👉 zpid.cloud.panopto.eu/Panopto/Page...

#TextAsData #Embeddings #LLM #NLP #Education
April 2, 2024 at 7:54 AM
ariedamuco.bsky.social
This scales corruption measurement to the universe of audited municipalities. It's transparent, replicable, & cost-effective (unlike hand-coding or LLMs).
📄 Paper: papers.ssrn.com/sol3/papers....
#TextAsData #Corruption #NLP #MachineLearning #Research #Methodology #OpenScience End/
Measuring Corruption from Text Data
Using Brazilian municipal audit reports, I construct an automated corruption index that combines a dictionary of audit irregularities with principal component a
papers.ssrn.com
December 9, 2025 at 1:12 PM
chrauh.bsky.social
➡️ Predoctoral Research Fellow / PhD Candidate
👉 75%, 3 years, @wzb.bsky.social
👉 Focus: Quantitative text analysis ( #NLP | #CSS | #TextAsData ) on global crisis perceptions and dissertation research
👉 Full job call: www.wzb.eu/en/jobs/pred...
Predoctoral Research Fellow – PhD candidate (f/m/x) (ID 364)
For its VARICRIS project at the research department Global Governance, the WZB Berlin Social Science Center is looking for A Predoctoral Research Fellow – PhD candidate (f/m/x) to be employed...
www.wzb.eu
October 22, 2025 at 2:40 PM
jurimarcucci.bsky.social
🚨 LLMs+research=beware of #lookahead #bias 🚨

On Jan 16, 2026, join the #AMLEDS webinar w/ @amanela.bsky.social (WashU) on Lookahead Bias with #LLMs!

How do we do credible #forecasting or #TextAsData analysis when models have seen the future?

🕚 11am ET-5pm CET

#AI #EconTwitter #EconSky #NLP #ML
AMLEDS webinar on Look-Ahead bias with LLMs: Measurement and Mitigation Techniques with Asaf Manela (Washington University in St. Louis) and Andreas Joseph (Bank of England) as moderator! Jan 16, 2026!
January 13, 2026 at 2:45 PM
ckreudersonnen.bsky.social
➡️ Predoctoral Research Fellow / PhD Candidate
👉 75%, 3 years, @WZB Berlin
👉 Focus: Quantitative text analysis ( #NLP | #CSS | #TextAsData ) on global crisis perceptions and dissertation research
👉 Full job call: www.wzb.eu/en/jobs/pred...
Predoctoral Research Fellow – PhD candidate (f/m/x) (ID 364)
For its VARICRIS project at the research department Global Governance, the WZB Berlin Social Science Center is looking for A Predoctoral Research Fellow – PhD candidate (f/m/x) to be employed...
www.wzb.eu
October 22, 2025 at 2:06 PM
gaborbekes.bsky.social
Suppose you have *3 hours* to tell lawyers about main ideas in #textanalytics, #TextAsData. Structured vs unstructured text, bag of words, stop-words, n-grams, word frequency, creating variables from words, dictionaries, sentiment analysis, NLP. 🎓📄💬
Anything else?
December 2, 2024 at 1:10 AM