Johannes B. Gruber
@jbgruber.bsky.social
Senior Researcher @gesis.org // Data Editor @polcommjournal.bsky.social
🔎 political communication (#polsky + #commsky) with text analysis and #rstats (#opendata + #openscience)
🌏 JohannesBGruber.eu
👨💻 research software github.com/JBGruber
🔎 political communication (#polsky + #commsky) with text analysis and #rstats (#opendata + #openscience)
🌏 JohannesBGruber.eu
👨💻 research software github.com/JBGruber
The package comes accompanied by version 2 of the preprint: osf.io/preprints/so...
I really appreciate the people who have cited the package in their research (github.com/JBGruber/tra...), but Google Scholar doesn't always pick it up, so maybe cite the paper instead?
I really appreciate the people who have cited the package in their research (github.com/JBGruber/tra...), but Google Scholar doesn't always pick it up, so maybe cite the paper instead?
November 24, 2025 at 8:50 PM
The package comes accompanied by version 2 of the preprint: osf.io/preprints/so...
I really appreciate the people who have cited the package in their research (github.com/JBGruber/tra...), but Google Scholar doesn't always pick it up, so maybe cite the paper instead?
I really appreciate the people who have cited the package in their research (github.com/JBGruber/tra...), but Google Scholar doesn't always pick it up, so maybe cite the paper instead?
My Friday afternoon...
November 14, 2025 at 1:41 PM
My Friday afternoon...
Very happy to update the {traktok} #rstats readme. After 1.5 years, you can finally search TikTok again without access to the Research API. It's slow and a bit clunky, but it works! Thanks, @michaelgoodier.bsky.social for the crucial hint!
November 11, 2025 at 8:38 AM
Very happy to update the {traktok} #rstats readme. After 1.5 years, you can finally search TikTok again without access to the Research API. It's slow and a bit clunky, but it works! Thanks, @michaelgoodier.bsky.social for the crucial hint!
One good thing about developing software is that you can keep your own needs in mind. Like when you can never remember your username and use it as the example value 😅 #rstats
October 16, 2025 at 2:58 PM
One good thing about developing software is that you can keep your own needs in mind. Like when you can never remember your username and use it as the example value 😅 #rstats
@sebstier.bsky.social at #MEDem Conf: computational research of democracy stands in the shoulders of the few enthusiasts who create datasets, software and infrastructure for it. How can we move forward? Short answer: more collaboration & sharing!
September 30, 2025 at 12:28 PM
@sebstier.bsky.social at #MEDem Conf: computational research of democracy stands in the shoulders of the few enthusiasts who create datasets, software and infrastructure for it. How can we move forward? Short answer: more collaboration & sharing!
@simonsaysnothin.bsky.social at #MEDem Conf: we need to integrate our efforts instead of researchers all building their own datasets and infrastructure. Couldn't agree more!
September 29, 2025 at 11:49 AM
@simonsaysnothin.bsky.social at #MEDem Conf: we need to integrate our efforts instead of researchers all building their own datasets and infrastructure. Couldn't agree more!
Just wanted to share this Google Scholar trick: I often have the problem that I want to find papers using certain computational methods, but specifically in my own field (for lit reviews).
You can do that by limiting the search to certain sources. My (imperfect) collection in the alt text.
You can do that by limiting the search to certain sources. My (imperfect) collection in the alt text.
September 22, 2025 at 8:03 AM
Just wanted to share this Google Scholar trick: I often have the problem that I want to find papers using certain computational methods, but specifically in my own field (for lit reviews).
You can do that by limiting the search to certain sources. My (imperfect) collection in the alt text.
You can do that by limiting the search to certain sources. My (imperfect) collection in the alt text.
Works fine for me. Maybe the user agent got blocked after too many tries?
September 21, 2025 at 5:02 PM
Works fine for me. Maybe the user agent got blocked after too many tries?
If you want to test the 120B models of gpt-oss, you can do so using the Ollama Turbo (a API version of Ollama hosted by the creators) from #rstats 📦 {rollama} (it worked with my account even though I didn't give them my credit card info)
jbgruber.github.io/rollama/arti...
jbgruber.github.io/rollama/arti...
August 25, 2025 at 7:47 PM
If you want to test the 120B models of gpt-oss, you can do so using the Ollama Turbo (a API version of Ollama hosted by the creators) from #rstats 📦 {rollama} (it worked with my account even though I didn't give them my credit card info)
jbgruber.github.io/rollama/arti...
jbgruber.github.io/rollama/arti...
When I ask the question that appears everywhere in the Ollama docs, it gives the most scientific answer so far, which is cool. Asked to dumb down the answer a bit, it didn't convince me. The idea with the marbles is good but half-baked.
August 6, 2025 at 1:12 PM
When I ask the question that appears everywhere in the Ollama docs, it gives the most scientific answer so far, which is cool. Asked to dumb down the answer a bit, it didn't convince me. The idea with the marbles is good but half-baked.
When I extract the result from the response the old-fashioned way (regex extract from the reply), I get sentiment values. And now, suddenly, gpt-oss does really well and is quite fast!
August 6, 2025 at 1:12 PM
When I extract the result from the response the old-fashioned way (regex extract from the reply), I get sentiment values. And now, suddenly, gpt-oss does really well and is quite fast!
First time I'm pulling a model from 'Open'AI to my computer to run locally. I know it's just a publicity stunt to undermine Anthropic's release of the new Claude yesterday. But curious nonetheless!
August 6, 2025 at 9:33 AM
First time I'm pulling a model from 'Open'AI to my computer to run locally. I know it's just a publicity stunt to undermine Anthropic's release of the new Claude yesterday. But curious nonetheless!
I talked to some people recently who said they got bored or overwhelmed by Bluesky after a while. Turns out they didn't know about alternative feeds yet and we're scrolling the chronological one.
If you like that, great, but here are some alternative feeds I like.
If you like that, great, but here are some alternative feeds I like.
July 30, 2025 at 1:58 PM
I talked to some people recently who said they got bored or overwhelmed by Bluesky after a while. Turns out they didn't know about alternative feeds yet and we're scrolling the chronological one.
If you like that, great, but here are some alternative feeds I like.
If you like that, great, but here are some alternative feeds I like.
I then tried two different approaches: 1. iterating through the accounts and searching videos where it EQuals the creator; 2. looking for videos where the creator is IN the list of party accounts
July 23, 2025 at 7:51 AM
I then tried two different approaches: 1. iterating through the accounts and searching videos where it EQuals the creator; 2. looking for videos where the creator is IN the list of party accounts
I wanted to see which videos the German parties/top candidates had posted between the announcement of the federal election and the election. These are the accounts I found in a 5 minute search
July 23, 2025 at 7:51 AM
I wanted to see which videos the German parties/top candidates had posted between the announcement of the federal election and the election. These are the accounts I found in a 5 minute search
I don't think anyone ICAed harder than @damiantrilling.net, at least judging by his badge. #ica25
June 16, 2025 at 10:44 PM
I don't think anyone ICAed harder than @damiantrilling.net, at least judging by his badge. #ica25
I would summarize this great last presentation of #ica25, but @camilambpp.bsky.social has already done it better 👇
June 16, 2025 at 10:06 PM
I would summarize this great last presentation of #ica25, but @camilambpp.bsky.social has already done it better 👇
Always worth sticking around until the end of #ica25: @profvaccari.bsky.social presenting an insightful study on whether people trust misinformation on WhatsApp simply because someone put a BBC logo on it (they do).
June 16, 2025 at 9:29 PM
Always worth sticking around until the end of #ica25: @profvaccari.bsky.social presenting an insightful study on whether people trust misinformation on WhatsApp simply because someone put a BBC logo on it (they do).
Some people have left #ica25 already, but great research is still being presented, like @gongbaobao.bsky.social who shows this co-consumption network of German media
June 16, 2025 at 8:32 PM
Some people have left #ica25 already, but great research is still being presented, like @gongbaobao.bsky.social who shows this co-consumption network of German media
Happening now #ica25: how high is the quality of the data you get from the TikTok research API? @mshu.bsky.social , me, @zwizwasvens.bsky.social, Johanna Klapproth, and @thorstenquandt.bsky.social have an answer: it's bad!
June 16, 2025 at 4:48 PM
Happening now #ica25: how high is the quality of the data you get from the TikTok research API? @mshu.bsky.social , me, @zwizwasvens.bsky.social, Johanna Klapproth, and @thorstenquandt.bsky.social have an answer: it's bad!
You want to study German covid-19 protests? @maxzehring.bsky.social, @hdschulze.bsky.social and @buehling.bsky.social got you covered with their dataset! #ica25
June 16, 2025 at 4:34 PM
You want to study German covid-19 protests? @maxzehring.bsky.social, @hdschulze.bsky.social and @buehling.bsky.social got you covered with their dataset! #ica25
For example, for this panel at 10:30 where @mshu.bsky.social will present our audit of the TikTok Research API and I will talk about non-consumptive research in #commsky (#ica25)
June 16, 2025 at 2:59 PM
For example, for this panel at 10:30 where @mshu.bsky.social will present our audit of the TikTok Research API and I will talk about non-consumptive research in #commsky (#ica25)