@ahalterman

NLP, event extraction, and political violence. Assistant professor of political science at MSU. PhD from MIT. Mastodon: @andy@sigmoid.social

 Ann Arbor, MI, USA
andrewhalterman.com
Joined June 2009
  • Tweets844
  • Following243
  • Followers921
74Media
Currently in FirstView: “Synthetically generated text for supervised text analysis.” @ahalterman proposes a method for using LLMs to generate synthetic training data for training smaller, traditional supervised text models.
1
15
46
5,594
Reminder - for the terrific interdisciplinary Text as Data conference, abstract submissions coming up - due Aug 4! tada2023.org/ It's a great, small, non-archival conference to discuss emerging work with folks across social sciences, humanities, and computer science.

TADA 2023: New Directions in Analyzing Text as Data

tada2023.org
40
2
66
26,360
MSU political science is hiring in methods! I'm at PolMeth today if you want chat about it. careers.msu.edu/en-us/job/51…
1
13
1
25
7,219
Or if you're interesting in forecasting civil war as a latent variable, you can come talk to me about that too.
1
439
How can we categorize political actors extracted from text without or dictionaries or lots of hand labeling? We can use a "soft dictionary" approach with a small set of hand-written patterns and a transformer model.
1
2
8
1,050
Blog post: andrewhalterman.com/post/sof… Live demo: andrewhalterman.com:9067/ Paper: arxiv.org/abs/2304.01331
1
4
521
OK, here it is: a line in the sand (in @Nature). I am very wary about scientists---including political scientists---embracing/pushing proprietary LLMs. Let's try an open science approach. Hope this take is a useful one. nature.com/articles/d41586-0…

Why open-source generative AI models are an ethical way forward for science

Nature - Researchers should avoid the lure of proprietary models and develop transparent large language models to ensure reproducibility.

nature.com
16
116
18
360
398,949
Big epiphanies today in the grad causal inference class reviewing all the methods we've covered so far.
The "always has been" astronaut meme, with one astronaut saying "The missing potential outcome is just filled in with a prediction?" and the other astronaut saying "Always has been"

ALT The "always has been" astronaut meme, with one astronaut saying "The missing potential outcome is just filled in with a prediction?" and the other astronaut saying "Always has been"

1
2
8
1,023
I've (finally) released an update to my text geoparsing library. Mordecai 3 lets you pass in a document, and returns the place names from the text and their geographic coordinates. It's built on #spacy and Geonames and uses a new neural similarity model.
A diagram of the main neural similarity model in Mordecai 3, showing comparisons between placename and document embeddings and entries in the Geonames gazetteer.

ALT A diagram of the main neural similarity model in Mordecai 3, showing comparisons between placename and document embeddings and entries in the Geonames gazetteer.

2
4
1
41
5,438
Mordecai 3 is open source, runs offline, and is available via pip or on GitHub: github.com/ahalterman/mordec…. Check out the paper for more details and performance comparisons with other geoparsers: arxiv.org/abs/2303.13675

GitHub - ahalterman/mordecai3: Full text geoparsing/toponym resolution with event geolocation

Full text geoparsing/toponym resolution with event geolocation - ahalterman/mordecai3

github.com
1
6
320
No ChatGPT here (yet), only the finest, handcrafted, artisanal features and good old RoBERTa.
7
316
Andy Halterman retweeted
cs.CL Papers@arxiv_cs_cl5 Apr 2023
ift.tt/4aNvbci Creating Custom Event Data Without Dictionaries: A Bag-of-Tricks. (arXiv:2304.01331v1 [cs.CL]) #NLProc
1
1
845
Andy Halterman retweeted
Andy Halterman@ahalterman28 Mar 2023
Replying to @emollick
Not a special issue, but I have a working paper on using synthetically generated text to train supervised classifiers (with poli sci applications). Paper: andrewhalterman.com/files/Ha… Poster: andrewhalterman.com/files/Ha…
1
1
1
201
Andy Halterman retweeted
Niklas Stoehr@niklas_stoehr8 Dec 2022
Replying to @ben_j_radford
@ben_j_radford and @hurrial kicking off the second day of the #CASE Workshop (Extraction of Socio-political Events from Text) @emnlpmeeting. @ahalterman @tiancheng_hu @yaoyao_dai @HristoTanev2 🙌 – at Abu Dhabi, United Arab Emirates
2
5
Andy Halterman retweeted
MIT SSP
@MIT_SSP1 Dec 2022
Much of IR is concerned with understanding the behavior of elites. That’s nice from a natural language processing perspective, as elite decision-making tend to get written down. — SSP alum @ahalterman on natural language processing in IR research. e-ir.info/2022/11/24/intervi…
1
2
3
If you’re in Ann Arbor on Friday, I’ll be talking about computational tools for political science research.
Center for Political Studies@umisrcps8 Nov 2022
Join us Friday at @umisr for a roundtable on mismatches between computational tools and social science needs. With Walter Mebane, @ahalterman @cjfariss and @jmendelsohn2.
Learning how to play with machines roundtable on Mismatches between computational tools and social science needs: with Walter Mebane, Andy Halterman, Chris Fariss, and Julia Mendelsohn. Moderated by Arya McCarthy, Giovanna Maria Dora Dore and Eva M. E. Klaus. Nov. 11 at 1:30 pm in ISR Room 1430.

ALT Learning how to play with machines roundtable on Mismatches between computational tools and social science needs: with Walter Mebane, Andy Halterman, Chris Fariss, and Julia Mendelsohn. Moderated by Arya McCarthy, Giovanna Maria Dora Dore and Eva M. E. Klaus. Nov. 11 at 1:30 pm in ISR Room 1430.

5
Have you (or someone you know) struggled to get coherent topics out of the LDA topic model? ✍️ I've written you a list of tips, compiling what I know from the research literature + my personal experience modeling many different datasets. maria-antoniak.github.io/202…

Maria Antoniak

My academic website / portfolio.

maria-antoniak.github.io
11
98
6
467
Modeling proportions/beta distributed DVs comes up all the time in political science and I’ve wanted something that easily handles them for so long!
Robert Kubinec@rmkubinec27 Jul 2022
🚨 Happy that my ordered beta regression article is now online @polanalysis! 🚨 Ordbetareg is a drop-in replacement for OLS when modeling bounded continuous outcomes--such as proportions, indexes, sliders. #rstats📦: cran.r-project.org/web/packa… Paper link: bit.ly/3czIUIc

ordbetareg: Ordered Beta Regression Models with 'brms'

Implements ordered beta regression models, which are for modeling continuous variables with upper and lower bounds, such as survey sliders, dose-response relationships and indexes. For more informa...

cran.r-project.org
1
4
Free app idea: scape the NYC scaffold permits and give rain-specific walking directions that minimize your uncovered walking time.
5
I've been getting unalloyed joy from @midjourney. I made some signs for the kegs at my wedding this weekend. (Prompts in the alt text)
A label for "Brookside Orchards Cider" generated with the prompt "an apple turns into a glass of cider, illustration, 19th century naturalist, schematic drawing"

ALT A label for "Brookside Orchards Cider" generated with the prompt "an apple turns into a glass of cider, illustration, 19th century naturalist, schematic drawing"

A label for "Maine Beer Company's Lunch IPA". Generated with prompt "a glass of beer containing a microcosm of a Maine seascape, illustration"

ALT A label for "Maine Beer Company's Lunch IPA". Generated with prompt "a glass of beer containing a microcosm of a Maine seascape, illustration"

1
9
Load more