Microblog

Replying to @b0rk

@b0rk Thanks for writing this! Small-business-owner high-five! I loved your first business blogpost and was hoping you'd written an update since Jan 2019. ^^ I very much recognize the 'start email list, then abandon ship' strategy, so it was useful to read about your experiences. :)

Sooooo The Pudding made a judgmental text generator for music lovers: https://pudding.cool/2020/12/judge-my-spotify/

I finished Alba: A Wildlife Adventure! I am a huge fan of the game and cannot recommend it enough. Buy it if you like relaxed indie games with nature, cute animals and in-game photography. Here are my favorite playthrough photos.

'Alba: a Wildlife Adventure' is everything I ever wanted in a video game. It's basically Pokemon Snap meets Windfall Island.

Last year, @lorgatti and I (and some of our @utwenteEN students) tried to improve the state of Dutch open-source sentiment analysis -- and our work is now published in CLIN journal! https://www.clinjournal.org/clinj/article/view/105

Building a new Datakami Labs #nlproc product today in tech-incubator Ploegstraat in Nijmegen, together with @Wassasin of http://woutergeraedts.nl and these scrumptious “koffiebroodjes”...

Replying to @Roberdus

@Roberdus Something for @CPNB? Or maybe they can bring you into contact with a local library, such as @biebenschede?

Three of my ANLP students are experimenting with the Star Wars: KOTOR dialogue dataset + BERT + natural language generation for video games. First results are fun and promising. Slides w/ prelim results are from the students, the blue remarks are mine.

<NLP> The state of the art in sentiment analysis uses deep learning! Long live neural models! Sentiment lexicons are dead!
<Industry> No they are not https://twitter.com/albrgr/status/1333838686992044032

Replying to @zehavoc

@zehavoc @albrgr I don't find it crazy at all either (but you know, I'm biased, I research NLP AND I'm interested in investing). Retweeted this tweet because it gives a nice overview of the news article.

Finished #NaNoWriMo with 15,113 words -- and I accidentally discovered my best hour in the day for thesis writing.

Hey leuke mensen van @PinguinRadio, ik heb (geautomatiseerd) aan Spotify gevraagd om een playlist te bakken van de Snob 2000 keuzelijst voor 2020. Voila: https://open.spotify.com/playlist/6prwQaz4stoMgTxa5LiiIt?si=Zm4vhNhFTbuOGCkL36wmrg

Replying to @jd7h

@PinguinRadio Spotify heeft 6575 van de 7227 liedjes gevonden. Er kunnen wat kleine foutjes inzitten; mijn script zoekt binnen Spotify-voor-Nederland op "[artiest] [titel]" en pakt dan de eerste hits van de zoekresultaten (if any).

It's time to throw a few prototypes into the world. Let's make NLP a bit more tangible. :D https://t.co/W7JWbfjVAC

LinkedIn, why have you wrongly profiled me as an anthropologist!? I know I'm a generalist, but this is taking it too far.

Newsletter: Inspiring entrepreneurs and the start of something beautiful - https://mailchi.mp/671dceaa7c39/inspiring-entrepreneurs-and-the-start-of-something-beautiful

Replying to @jd7h

In which I talk about my favorite business owners @honnibal, @_inesmontani, @b0rk and Mark Vletter (@VoysNL).

My alma mater @Radboud_Uni is hiring a senior Game Developer for the Games for Emotional and Mental Health (GEMH) Lab. https://solliciteren.ru.nl/cgi-bin/share/og_pxs.pl?target=LI&recid=1131492

Replying to @peachhugs

@peachhugs Thanks! It definitely was one of the most fun projects of my PhD, together with extracting text from popular video games.

Replying to @peachhugs

@peachhugs You can find the updated version of the paper here: https://judithvanstegeren.com/assets/vanstegeren2020fantastic.pdf :)

@chrisamaphone Are the slides for your INT2020 talk available somewhere (or a recording of the talk)?

Replying to @jd7h

@chrisamaphone Found them! https://docs.google.com/presentation/d/1AjsRbQHUqMSPuf0ZEJa2g-EnAhIaKopLGSEisvznTGo/edit?usp=sharing

KLABAM, passed the 10,000 thesis words! I've never edited my #NaNoWriMo draft before, so I'm curious to see what'll happen to the draft over the next few months.

This is a screenshot of my macro-economics Grafana dashboard, which contains data from both Dutch and US sources. Interesting to see how the CNN Fear & Greed index was influenced by the US elections.

I'm building a text generator that can write financial analysis reports for value investors. It's so much fun when multiple interests (data, Pandas, text generation, investing) come together in one project!

This piece about Digital Gardens makes me so happy. We need more wacky, out-of-the-box websites! https://www.technologyreview.com/2020/09/03/1007716/digital-gardens-let-you-cultivate-your-own-little-bit-of-the-internet/

Next week the advanced research course on natural language processing starts again. It's always great to see what students come up with (new datasets, techniques) and sometimes the course even leads to a publication. 4 of these topics are mine, can you guess which ones? ;)

Replying to @jd7h

The answer to my riddle! The other topics are from my wonderful NLP colleagues at @HMIutwente including @jtvwts, @lorgatti and @shenghui. :)

Hit the 5000 word mark today!
$ Apply(Seat(pants),Seat(chair))
#NaNoWriMo

My dissertation research turned out to be mostly about continuous delivery, integration and scheduling algorithms for coffee-to-text generators in an academic environment.

After searching for other NaNoWriMo tweets: prepare for a barrage of Hamilton memes https://twitter.com/jd7h/status/1321585737180106757

I’m joining #NaNoWriMo this year as a first time NaNo rebel: I’ll write 500 words a day for my PhD thesis, so that’s 15,000 words in November. Bring on the caffeine, frozen meals and online Write-Ins! Who’s with me?

Wow, people tried to use GPT-3 for partially automating the work of a GP!? That's a horrible idea for a lot of reasons. "Putting GP in GPT-3" 🤣

https://www.theregister.com/2020/10/28/gpt3_medicine_ai/

Reading material for #phdlife: this is a pretty good resource about imposter syndrome, especially the tips at the bottom of the graphic. https://resume.io/blog/are-you-suffering-from-impostor-syndrome

A highlight of @AIIDEconference so far is the work of @rajammanabrolu et al. on story world generation (including flavour text). Smart use of both GPT-2 and BERT in the architecture, and the authors have conducted a solid evaluation with actual games. Cheers!

Replying to @jd7h

@AIIDEconference @rajammanabrolu Paper here: https://www.aaai.org/ojs/index.php/AIIDE/article/view/7400/7309

Presented 3 new datasets with video game text at the INT2020 workshop at @AIIDEconference!

YouTube video: https://www.youtube.com/watch?v=hBSTR_g0btg

Paper: https://judithvanstegeren.com/assets/vanstegeren2020fantastic.pdf

Datasets (with spoilers for KOTOR1 and Torchlight II): https://github.com/hmi-utwente/video-game-text-corpora

Replying to @KimutaiGibson

@KimutaiGibson @AcademicChatter @PhDVoice Celebrating when you submit the paper instead of celebrating when a paper gets accepted makes a huge difference for me. You often can't control whether the reviewers will like your project/methods/research focus, but you can control the quality of your research and writing.

Replying to @jd7h

@KimutaiGibson @AcademicChatter @PhDVoice If my projects and papers align with my own scientific values, work ethic, and ideas about project usefulness and paper readability, I'm not so anxious for the peer-reviewing process.

I just received my copy of Cosy Days: The Art of Iraville, by @Iraville_de. The illustrations are amazing and perfect for autumn. A flip-through video (not by me) can be found here: https://www.youtube.com/watch?v=lHQ_bw-rhRg

Prof Lucy Rogers PhD FREng @DrLucyRogers

This simple trick for 30 seconds every day really does make a difference (speak to your camera and WATCH it back).

Can confirm! My personal implementation was 'do a 2-hour Twitch stream every week for 6 months... and watch it back', which is a bit more labor-intensive. Learned a lot of cool stuff about speaking/presenting/non-verbal communication and A/V tech! https://twitter.com/DrLucyRogers/status/1317034203624734722

Sometimes I read papers about topics that are adjacent to my own research. This is an interesting one: authors evaluated web search engines (including a few obscure ones) as a linguistics research tool. https://www.sciencedirect.com/science/article/pii/S0306437917303290

Tom Gauld @tomgauld

#research A recent cartoon for @newscientist.
p.s. My book of science cartoons is out now: https://www.tomgauld.com/

This could be one of those "What my parents think I do, what my boss thinks I do, what I really do" memes. https://twitter.com/tomgauld/status/1315637508697456641

Replying to @BartEnkelaar

@BartEnkelaar Ik ben opgegroeid met (een gecensureerde, kind-vriendelijke subset van) Drs. P compilé sur CD, dus ik hoor het hem gewoon zingen. :)

Wanneer u eens, al is 't maar voor de grap
Een druppel water haalt uit de Zambezi
(Het mag ook uit de kraan zijn, of uit de Rijn)
En deze neerlegt op een multomap
Een koekblik, of een briefkaart uit Berlijn
Behoudt hij zijn gestalte door cohesie
Dat is een waardevolle eigenschap

Replying to @jd7h

Gedicht "Druppel" van Drs. P. (Een serieuze kandidaat voor pagina 1 van mijn proefschrift)

Gisteren in het NRC: een interview met Antal van den Bosch, directeur van het Meertens Instituut. ‘Vertaalprogramma werkt zoals je brein’ https://www.nrc.nl/nieuws/2020/10/04/vertaalprogramma-werkt-zoals-je-brein-a4014623

Replying to @drtowerstein

@drtowerstein Recipe here: https://www.pickuplimes.com/single-post/2019/11/05/Chai-Hot-Chocolate

I love autumn. 🍂🍁 Time for some hot chocolate chai!

Replying to @joynessthebrave

@joynessthebrave Congratulations on the submission!

This is a helpful thread for preparing my talk at #INT2020 @AIIDEconference. https://twitter.com/annetropy/status/1305240409686380552

New blogpost: I've posted my list of computer security learning resources to my blog. https://judithvanstegeren.com/blog/2020/computer-security-resources-for-self-study.html

@Honeydew_Mellen Are there any plans for your lovely ghost illustrations, for example to bundle them into a book at some point? They're my favorites. ^^

Ellen Stubbings @Honeydew_Mellen

Day 4 of #helloautumnweek - autumn picnic 🍂

Art like this is one of the reasons I'm still on Twitter https://twitter.com/Honeydew_Mellen/status/1309213330339438592

Replying to @JurieHorneman

@JurieHorneman Haha, dank je! Perfect responsive, doet het in iedere browser, supersnel en vereist weinig energie. Wat wil een mens nog meer?

I've officially registered a name for my consultancy business: Datakami! \o/ I'm available for projects where NLP and data engineering meet. If you want to know more, http://datakami.nl has my contact details. :)

Replying to @jd7h

...where NLP stands for Natural Language Processing.

Replying to @onzetaal

@onzetaal Wat ik wel normaal vind klinken: 'Met 16 (jaren) mocht ik alleen naar een festival', als variant op 'toen ik 16 was' of 'op m'n zestiende'

New blogpost: https://judithvanstegeren.com/blog/2020/financial-data-sources-for-individual-investors.html

@AIIDEconference Any updates on this? The answers really influence my registration decision as I'll be following the conference/workshop from Europe. https://twitter.com/FloRicx/status/1301670063213215744

I was fine-tuning GPT-2 on a small test-dataset (the top n lines of a big sorted file), and now I have accidentally learned it that headlines should always start with a capital A. My fine-tuned GPT-2 thinks Argentina, Afghanistan and Austria are the only countries in the world.

Tom Gauld @tomgauld

‘How to tell if your cat is interested in the novel you are writing’ (my cartoon for yesterday’s @guardianreview)

s/novel/dissertation #phdlife https://twitter.com/tomgauld/status/1305069068421156866

In funny bugs: Python's datetime accepts dates starting in 1 CE, and consequently, my book tracking program has problems with the book Tao Te Ching, which has a 'first published' date somewhere in the 4th century BCE.

@tomcritchlow For your bookshelves-list: I admire http://books.hansdezwart.nl. Since it's built in PHP and not open-source I'm trying to build something similar in Python/Django: https://github.com/jd7h/josephine/

So this proposal for decentralized bookshelves-on-the-web lead me to a whole world of new side-project crack... https://tomcritchlow.com/2020/04/15/library-json/

Why Goodreads is bad for books https://www.newstatesman.com/science-tech/social-media/2020/08/better-goodreads-possible-bad-for-books-storygraph-amazon

This research paper describes how PhD examiners (thesis committee) read PhD theses, based on 30 interviews with experienced examiners. Interesting material for those living the #phdlife.
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.187.6401&rep=rep1&type=pdf

@ptychomancer I'm investigating the 'apophenia' phenomenon, inspired by your work on the biographies in Caves of Qud. Is there any literature I should absolutely read?

TIL git rebase
https://christoph-rumpel.com/2015/05/clean-up-your-commits-for-a-pull-request/

Bert Hubert 🇺🇦 @bert_hu_bert

Replying to @bert_hu_bert

Live in 60 minutes on https://nlnog.net/webcast/ !

Preach! 👏 https://twitter.com/bert_hu_bert/status/1301824202404700163

That moment when your whole life makes sense and everything just falls into place: I just used my security education (tcpdump) to debug the activeMQ part of a new text generation system. \o/

Great news for privacy-aware people and NLP researchers!
Companies are slowly (re)discovering the advantages of online ads that are not personalised (based on tracking cookies and user profiles) but contextualised (based on the content of the webpage).

Replying to @jd7h

According to http://FD.nl "an algorithm distills tv programme subtitles into 23 different context labels, like 'sports', 'travel' or 'culture'. Based on these labels, ads get a specific location (on the website of NPO, the Dutch public broadcasting organisation)."

Replying to @jd7h

Source (in Dutch): https://archive.is/hxtkF

Replying to @zehavoc

@zehavoc https://twitter.com/jd7h/status/1298979060001861633

Preprint: https://judithvanstegeren.com/assets/2008-vanstegeren2020fantastic-preprint.pdf
Datasets: https://github.com/hmi-utwente/video-game-text-corpora https://twitter.com/jd7h/status/1298535577688694784

Replying to @zehavoc

@zehavoc Let me doublecheck the rules of the conference, and I'll get back to you :) I've never put a preprint up before publication because all my papers so far are published as open access papers.

Replying to @xfoml

@xfoml Thanks for suggesting this! I ended up extracting texts directly from game files with modding tools. The paper was not ready in time for LREC, but it will now be published at the Intelligent Narratives workshop at AIIDE 2020: https://twitter.com/jd7h/status/1298535577688694784

Happy to announce that I got a paper accepted at INT20 (workshop at AIIDE 2020), in which I describe three corpora with game texts from commercial video games TorchLight II, SW:KOTOR and The Elder Scrolls! I hope it will benefit researchers in NLP, PCG and video games & AI.

Replying to @joynessthebrave

@joynessthebrave Thanks, that makes sense.
I'd love blogposts/podcasts about this, with (feelgood) books, music and other recommendations, since I'm really enjoying your tweets and writings so far. I've started listening to Vienna Teng thanks to you. ^^

@joynessthebrave Starting The Dissertation soon. Do you have any resources to recommend for keeping your sanity in the final year of your PhD?

I want to write a short email course about something NLP/data science related. Which topic would you be most interested in?

PSA: If you like books, reading, illustrations, special editions, rare editions, rare special illustrated editions or hoarding, stay away from books titled 'Book Collecting: A Guide'.

Just blogged: My data-driven investment toolkit https://judithvanstegeren.com/blog/2020/data-driven-investing-toolkit-hetty.html

@Superflux Where can I get a copy of Instant Archetypes? Your Shopify-shop is unavailable. I think the deck would be a perfect interdisciplinary brainstorm tool for my research group.

Replying to @drtowerstein

@drtowerstein @me_datapoint @eScienceCenter Sorry I forgot to post the link! My brain's not functioning at 100% -- it must be because of the heat wave. ;)

Apparently, NLP's favorite example for "additive compositionality" of word vectors, i.e. "king - man + woman = queen", works in a lot of cases only if you cheat by excluding the original word from the vector space! Interesting article by @me_datapoint of the @eScienceCenter!

Replying to @jd7h

@me_datapoint @eScienceCenter Article here: https://blog.esciencecenter.nl/king-man-woman-king-9a7fd2935a85

A very interesting essay about the fields that constitute "data science", by prof Jeffrey D. Ullman. http://sites.computer.org/debull/A20june/p8.pdf

Want to use a state of the art language model, but can't choose between GPT-2 and BERT? I had the same question. I did some reading up on both models and created an overview based my findings: https://judithvanstegeren.com/blog/2020/GPT2-and-BERT-a-comparison.html #NLProc #nlp #nlg

This goes for websites, but also blockchain, NLP and digital security systems.
"Digital is greedy for energy and the more it grows the greedier it gets. We need digital designers who think about the weight of every design decision they make." https://alistapart.com/article/webwaste/

Good news for people researching tabletop games and RPGs: researchers have published an annotated version of the collection of @CriticalRole episode transcripts (thanks, #critters!). Now we can combine linguistics research with our favorite D&D show. https://www.aclweb.org/anthology/2020.acl-main.459/

My favorite keynote at IEEE COG 2019, "Natural language in games" by @emshort, was uploaded to YouTube. It covers natural language processing, understanding AND generation.
https://www.youtube.com/watch?v=iBCmwvLxp24

This concept of "ventilated prose" could come in handy when I'm (re)writing papers and thesis chapters... https://vanemden.wordpress.com/2009/01/01/ventilated-prose/

Dear friends! I've created multiple datasets for fellow researchers this year, and I've found that *data wrangling* is something I love doing. I'd like to take on a few consultancy projects this year, so let me know if I can help! More info: https://judithvanstegeren.com/consultancy/

Interesting analysis of how (moral) choice systems are implemented and communicated to the player in RPG video games: https://withaterriblefate.com/2016/04/25/beyond-the-moral-binary-decision-making-in-video-games/

Security people and AI people should work together more often.

Replying to @jd7h

AI people can bring in their magic to summarize and filter the heaps of security data; security people can bring in their distrust of untrusted, third-party (training) data and hopefully mitigate some of the biases in AI systems.

Tom Gauld @tomgauld

#libraries

Readeasies is my new favourite word. https://twitter.com/tomgauld/status/1282242816824025088

harley turan @hturan

Here's GPT-3 generating React components based on a _variable name_ alone, using the @OpenAI API.

Now I wonder whether the training set for GPT-3 also includes code from the Underhanded C Contest... What happens if your AI starts generating obfuscated malicious code? https://twitter.com/hturan/status/1282261783147958272

Replying to @myrthereuver

@myrthereuver @antske @VUamsterdam @suzan @UniLeiden Congrats!

Faierius' Fables & Flossworks @FaieFableFloss

Grammarly, what in sweet baby jeebus is this?

Hilarious thread. Also: don't train your AI systems on untrusted input. https://twitter.com/FaieFableFloss/status/1281084519311765504

Dutch newspaper het Financieele Dagblad has a cool long-read today about how architects use generative design for exploring different options during the design phase of a project.
http://archive.is/w8ect

Inspiring essay by @mattmight about (the road to) tenure and how to deal with all the related conventional 'wisdom': http://matt.might.net/articles/tenure/

Researchers developed a method for predicting cryptocurrency prices with sentiment analysis and published their results in a journal. When they got interviewed by a Dutch newspaper they stated "Don't worry, our method can't be copied. We left out essential details in the paper."

Replying to @jd7h

How is this science? It makes me a bit sad.

Replying to @jd7h

Dutch news source (article in Het Financieele Dagblad): https://archive.is/riny4

For those who missed it: GPT-3 is here! https://www.theguardian.com/technology/2020/jun/12/elon-musk-backed-openai-to-release-text-tool-it-called-dangerous

Reading 'An Eames Anthology': the writings of Charles and Ray Eames, the designers/architects. Computer scientists could learn a lot from the philosophy and approach of these artists.
What if we started to build AI systems this way? (I'm not convinced we are doing this now)

An here's another piece of evidence that sometimes, the coolest research on procedural text generation for games happens outside of (the official channels of) academia: https://www.vice.com/en_us/article/bmjke3/the-ai-that-learned-magic-the-gathering

Replying to @SketchesbyBoze

@SketchesbyBoze The Dream Merchant by Isabel Hoving.

By popular demand: tonight at 20:00 (Amsterdam time), @mrngm and I will play The Long Dark story mode on Twitch! Join the stream at https://www.twitch.tv/jd7h

Replying to @jd7h

Unfortunately, we have some trouble getting a good quality stream. Currently game + OBS + twitch stream is too heavy for the computer. Sorry folks, we'll try to fix it but it means no stream tonight. (CC @drtowerstein)

#PCG people, I'm looking for blog posts or papers about best practices for creating and releasing a dataset in our field! Asking for the greater good of mankind, art and science ;)

Replying to @drtowerstein

@drtowerstein Ok! I've scheduled a Twitch stream for this Thursday, 20:00 (the usual time).

The Long Dark, aka post-apocalyptic camping simulator, is currently my favorite corona-proof holiday activity. My partner & I take turns playing and telling the other person to watch out for wolves, keep track of our fire wood, FIFO the food stash and maintain our socks!

The proceedings for the Games and NLP (#GAMNLP) workshop at LREC 2020 have been published! Proceedings can be found here: https://lrec2020.lrec-conf.org/media/proceedings/Workshops/Books/Games-NLPbook.pdf

Replying to @jd7h

GAMNLP accepted my joint work with Mariët Theune and Therese Bergsma on adapting a sentiment analysis lexicon to text from The Elder Scrolls games! (First paper in the proceedings pdf)

monkeyUser @ismonkeyuser

Caption this. Let’s have a laugh -🙈🙉🙊⁣⁣⁣- #CaptionThis

"We're looking for someone from Operations who can join our sales meeting." https://twitter.com/ismonkeyuser/status/1255754397272989696

@bunq Wat schandalig dat jullie mijn subscription-prijs onaangekondigd met 800% verhogen, en dat verpakken als heugelijk nieuws in een mailing. "Hoera, we hebben nieuwe subscription-modellen!" en "Hoera, je krijgt 3 maanden gratis Premium!".

Replying to @BuurtaalMiet

@webredMiet @onzetaal Als je een baan hebt waarbij dat de norm is, noem ik dat 'remote werken'. Werken op een andere plek voor een gewone Nederlandse baan waarvoor je normaal gesproken toegang tot een kantoor hebt, noem ik eigenlijk altijd thuiswerken, zelfs als ik niet thuis ben.

I had some time off last week. It was more hectic than anticipated due to COVID-19, but I spent some time:
- streaming with OBS w/ @mrngm and @bwesterb
- making procedural art in Processing
- making procedural art in Python (MatPlotLib) https://twitter.com/mrngm/status/1239300390832799745

Public domain radio broadcasted a real treasure tonight. I was on-stream exploring a crypt filled with a spooky voice, skeletons, multiple dead mages and a lich. And then THIS song came on -- great timing! https://www.youtube.com/watch?v=QDs0NWaXN2I

Well played, GPT-2, well played.

@Honeydew_Mellen I'd love to get your Ghost Gardener illustration on a white t-shirt. Is that an option? :)

@tripofmice Awesome! I'm building something similar (but different): https://github.com/jd7h/josephine

Fascinating list, both for researchers and lovers of sci-fi: https://100jobsofthefuture.com/browse/

Replying to @jd7h

Note: some of these job titles already exist. Others contain mostly buzzwords. If you decide to call yourself a blockchain talent analyst, I will laugh. Loudly.

The CFP deadline for the Games and NLP workshop has been extended to February 21st.
https://sites.google.com/view/gamnlp2020/

"A hair cut is a hair enhanced!" https://www.youtube.com/watch?v=ayzvXhTRbII
I am convinced I could finish my PhD by exclusively doing research inspired by Fry & Laurie sketches.

You know what made my Twitter experience LOADS better? I added a column to Tweetdeck with only images from #Astrophotography (with >5 likes and retweets). It reminds me daily that
(1) we are tiny carbon specks in space
(2) time passes
(3) the universe is awesome and beautiful

I love this write up about generating bookcovers by NYPL Labs: https://www.nypl.org/blog/2014/09/03/generative-ebook-covers #pcg

Feeling very sad about the state of academia today. A Dutch workgroup of academic staff recently published this report about work pressure: https://www.dropbox.com/s/x0ho46ivujq4kyf/WOinActie%20Survey%20of%20the%20Extent%20and%20Effects%20of%20Structural%20Overtime%20at%20Dutch%20Universities.pdf?dl=0.
My supervisor discussed the report with the local student newspaper: TIL she works 50+ hours w/a 32 hour contract.

Two of my interests combined in one video: @romanmars does live radio/podcast production in a TED Talk about design. Awesome! https://www.youtube.com/watch?v=pnv5iKB2hl4

@GET_TUDA_CHOPPA Very relatable, especially since all the games related conferences have their deadlines somwhere in feb and march? It took me some time to accept that there are not enough hours in the day to write a paper for each of them. Good luck with doing all the things. :)

"The first programmer is generally considered to be Adam Lovelace, Babbage’s collaborator and son of Lord Byron." 😂

https://billwadge.wordpress.com/2016/01/30/the-truth-about-adam-lovelace/

This satirical piece on programming approaches has aged well. :') https://billwadge.wordpress.com/2020/02/06/honest-wes-just-funnin-serious-work-viciously-parodied/

Replying to @_dmh

@_dmh @Dilbert_Daily Preach! 👏

s/slide deck/scientific poster/g 😭

https://dilbert.com/strip/2020-02-06 via @Dilbert_Daily

Replying to @myrthereuver

@rehtrym Did you try academictransfer? On that website you can set alerts for specifically PhD positions.

Replying to @drtowerstein

@drtowerstein Given the amount of research about Skyrim and the number of Elder Scrolls related datasets already for download, I think Bethesa doesn't mind.

Tom Gauld @tomgauld

This gif shows a few combinations of the artworks I made for the Cystic Fibrosis Unit at Bart's Hospital. Read more about it here: https://bit.ly/3bbPmjI
(thanks @Heart_Agency, @VitalArts, @BartsHospital)

Digitalized analog #PCG art by @tomgauld. https://twitter.com/tomgauld/status/1224396002406338563

Replying to @drtowerstein

@drtowerstein Sure! But I'd put them online so that everyone can use them. :P

Replying to @mtrc

@mtrc This is great, Mike! I just had a student work on domain-specific sentiment analysis for Skyrim, but I'd like to validate her method on a dataset from a different game. This one might be suitable.

And another addition to the video game corpora list: all text from RPG Disco Elysium. 🤩 https://judithvanstegeren.com/blog/2019/video-game-corpora.html https://twitter.com/mtrc/status/1224363284956110849

Replying to @xfoml

@xfoml I haven't read many papers focusing on resources so what would be the added value of a paper over a blogpost? What could the LREC audience get out of it? I'm asking since most of these are just plain text files.

Replying to @mtrc

@mtrc What, this is all text from Disco Elysium?! Awesome, I'm adding that to the list. <3 (I won't work with this dataset yet, because I still want to play the game.)

I've added a new dataset to my list of video game corpora. One of my students found out that http://Thuum.org includes a text-dump of Skyrim in their library. Useful when you want to do NLP on game texts. :) https://judithvanstegeren.com/blog/2019/video-game-corpora.html

Replying to @jd7h

@jd7h This might also be of use: https://gist.github.com/efonte/ce0b3a8f2651d2263d7085b2121d9f6c

Replying to @TheWritersGuild

@TheWritersGuild I am researching games & narrative as a PhD student and I'm having a hard time finding useful info about game writers & industry practices! Could the collected data from the survey be shared as well (in anonymized form)? This could greatly benefit games researchers. :)

Those were the days... https://eev.ee/blog/2020/02/01/old-css-new-css/

Proudly hacking away in Django for my self-hosted, open-source reading tracker Josephine. Newest feature: I can now use the @openlibrary API to get cover images.

@Pinboard The link 'hardware behind the site' on https://pinboard.in/about/ is broken.

Although most of them don't seem to realise it, researchers that write (or maintain or store or execute) academic code are ops too.
https://www.netmeister.org/blog/ops-lessons.html

Replying to @zoombapup

@zoombapup Procedural content generation would be a great topic to combine with letting students practice in finding good resources. It's def a game dev frontier and there are a lot of good resources out there.

Replying to @zoombapup

@zoombapup I'm also interested to know what you mean with 'I want to engage students with their own practice more'?

Replying to @zoombapup

@zoombapup Sure, they won't be too thrilled but it's great practice for post-university work -- both academia and industry jobs sometimes requires staff to give long talks on a few days notice.

Replying to @jd7h

@zoombapup Still, there's multiple things you can do:
- reserve the first few lectures for prep time (either unschedule them or organize guest lectures)
- make groups of two students and let them share the responsibility >>

Replying to @jd7h

@zoombapup - give students the first half of each lecture only, and reserve the second half so you can fix problems with the content yourself if necessary

Replying to @jd7h

@zoombapup - it also helps if you know the background from your students well. You can give them more responsibility if this is on a topic they are already a bit familiar with; if the topic is completely new, scale down the assignment or give more guidance.

I've written a blogpost with course formats you can use for research seminar courses at a university, based on master courses I followed around 2014. I hope this is helpful to university staff out there! 🙏 https://judithvanstegeren.com/blog/2020/research-seminar-formats.html #askacademia #phdlife

Replying to @jd7h

@xotoxot I included your format for privacy seminar (as I remember it from way back). :)

Replying to @Marjolijstje

@Marjolijstje @fd0_nl Is dit een OPSEC test? ;)

My newest blogpost contains a general audience abstract for my Alice & Eve 2020 poster: https://judithvanstegeren.com/blog/2020/alice-eve-nanogenmo-narrative-generation-poster.html

To people who say “there’s just no women in computing”: you’re wrong! We’re right here, at Alice & Eve 2020 @UTwente. If you want to work with us, come and find us. INVITE US to give talks, present posters, discuss interesting questions. We exist.

Replying to @jd7h

We work in quantum computing, bio informatics, mathematics, language tech, systems engineering, machine learning, physics, computer security, game development and human media interaction — as researchers, engineers, designers, managers and CEOs.

CEO @emilyjacometti is giving a great talk @UTwente about why we need more female founders. 👏 PREACH! 👏 #alice2020

I’m at the Alice & Eve 2020 symposium @utwente today. I’ll present a poster (#16) about my NaNoGenMo project during the poster session. Come say hi if you’re interested in story generation or PCG! #alice2020

I'll start streaming Star Wars: Knights of the Old Republic (2003) in a few minutes. Join the stream: https://www.twitch.tv/jd7h

Replying to @Autumnsburg

@Autumnsburg Using PCG moves the task from the asset creator to the programmer -- and it really depends on the resulting generator whether it will save time/money/resources.

Replying to @jd7h

@Autumnsburg For very specific domains/asset types we can say "yes", for some assets it's "sometimes" and for most the answer is "not really but we're trying out best to remedy that".

Reading another #PCG survey from Hendrikx et al. (2013) Although I have some problems with their taxonomy of game assets (it seems heavily biased towards RPGs/action games) it includes useful stuff: a list of commercial games with PCG, and some thoughts on generator tradeoffs.

Replying to @jd7h

It also contains references to evidence for why we want to use PCG in the first place (e.g. the explosion of amount of staff needed to produce a commercial game) -- something that is left implicit in more recent papers.

Replying to @jd7h

To clarify: I would like to see better evidence & reasons because in reality most commercial games are not using PCG at all. This discrepancy leaves me unsatisfied when I read "we made this generator because PCG relieves game devs of the authorial burden".

Amazon says I am not allowed to read over 9000 books in 2020. :(

Replying to @GalaxyKate

@GalaxyKate Why wouldn't you be allowed to choose... and choose again? To stay with the vertigo metaphor: you don't have to jump: there are stairs, and you are allowed to rest (and reflect and change course or destination) on every floor you pass on the way down. :)

Replying to @togelius

@togelius That's sounds like an interesting approach. I'm not working with neural techniques at the moment, but even if we can translate low-cost (rule-based/statistics based) approaches to the search-based PCG framework, that might be a win for game dev.

Replying to @fonolog

@fonolog Helemaal zelfstandig leuke (hoe definieer je dat?) dingen schrijven is inderdaad moeilijk voor computers, maar uit mens-computer-samenwerkingsverbanden komen interessante en creatieve dingen! Science fiction verhalen, game narratives, romans...

I'm reading the (classic?) search-based PCG survey paper by @togelius et al. (2011) and now I wish there were similar methods for specifically procedural /text/ for games. I love this approach but it does not seem to be a thing in creative language generation land?

🌿alex dingley🌿 @ f05 mcm london!! @nooskadraws

a #DnD illustration commission for @jd7h of their beautiful halfling monk Iza brewing some loose leaf tea 🌿🌌🍵

I found @nooskadraws’s work through the #criticalrole art gallery and def recommend working with him. He’s currently taking commissions so reach out if you’re looking for kickass art of your own PC’s! Expect good communication and more than fair rates for a full illustration. https://twitter.com/nooskadraws/status/1218489098597490688

Replying to @Kyle_J_Morgan

@Kyle_J_Morgan Bots: check out https://tracery.io
Nanogenmo: read other people's approach & code & output over at Github: https://github.com/nanogenmo/2019

Replying to @drtowerstein

@drtowerstein They are currently only broadcasted as livestream, not as video on demand. I do have copies of every session, and I might make them available later on. :)

Tonight at 20:00 (UTC+1) I'll play the puzzle game Hidden Folks. I've watched the trailer and it looks (and sounds) great: everything is handdrawn and ALL game sounds are performed by voice actors. Come watch and hang out at https://twitch.tv/jd7h. :)

I'm looking for literature about how readers (players) interpret procedurally generated text. Envisioned application: content generation for games.
Objective: generating text that appears to fit a larger narrative (e.g. the setting of a game). Any recommendations? #pcg #nlproc

Replying to @jd7h

It's hard to find the right keywords for lit search since it's a multidisciplinary problem (game studies, NLG/NLP, PCG, CC, cognitive science, humanities) and each field uses different terminology! And some words, like "coherence", are overloaded or not well-defined.

Replying to @jd7h

@nickmofo @mmateas @mtrc @maxkreminski @isaackarth @mark_riedl Any thoughts?

I figured that if I want to build useful tools for game writers, I better start learning their language. #gamewriting

Replying to @jd7h

And surprise! Researchers and game writers mean radically different things when they use words like narrative, story and genre.

Event staff <3 https://twitter.com/QuinnyPig/status/1215712220757118976

Replying to @matthewmercer

@matthewmercer Dear Matt, just remember that for every less-than-positive internet person, there are hundred of us that started contemplating playing or DM'ing D&D because of Critical Role. You just don't hear from us, because we're too busy dreaming of our own new campaigns and characters. <3

JAMES CROPPER plc │ Est 1845 @jamescropper

Artist creates an Endless Journey in paper with 479,001,600 unique combinations
Using James Cropper’s #Vanguard paper in Pink & Ivory colours, @tomgauld's Endless Journey myriorama was exhibited alongside rare originals from 19th century at The Shandy Hall
https://www.jamescropper.com/partner-stories/every-picture-tells-a-story

Combinatorial procedural content generation... in analog form! :D #pcg #procjam https://twitter.com/jamescropper/status/1216744789972070400

Replying to @xfoml

@xfoml Thanks! :)

@executivegoth @Marisha_Ray I was wondering whether you have read The Graveyard Book by Neil Gaiman? I'm reading the graphic novel adaptation (vol. 1) right now and I think it's something you might like!

New blogpost: an overview of my research and creative projects in 2019! https://judithvanstegeren.com/blog/2020/overview-2019.html

Replying to @Rose_D_Luna

@_Astro_Nerd_ I do career planning all the time, without expecting my opinion/plan to stay the same throughout my PhD. I recently read The Professor Is In, which I do recommend -- although the book is really cynical about academia. ;)

Books about art & artists have helped me so much with my PhD research in the past two years. We need books like The Artist's Way (Cameron) and Show Your Work (@austinkleon) as much as How To Write A Better Thesis (Evans) and How To Get A PhD (Phillips). #phdlife

I'm planning to evaluate text generator Churnalist by letting humans judge its outputs, so today I'm diving into the world of crowdsourcing & data collection platforms. Does anyone want to share their experiences?

🔎Julia Evans🔍 @b0rk

in march 2018, i started talking to @nostarch press about publishing some of my zines as a box set.

now it's finished and you can buy a box set of my first 7 zines here! https://nostarch.com/linuxtoolbox. (and you can get 20% off with the code WIZARD :))

Congrats to @b0rk and @nostarch for publishing this beautiful boxset. I'm adding this to my end-of-year book haul! Buy if you like:
- linux command line tools
- awesome cover illustrations
- writing by @b0rk
- rainbows https://twitter.com/b0rk/status/1174538663545528320

Replying to @jd7h

@b0rk @nostarch You can read some of them for free here: https://wizardzines.com/. Just scrolls down and click the black & white covers.