Microblog

Meet my research assistant. He is green and cuddly and a very good listener. I haven’t named him yet. #phdlife

If you're not familiar with the practices of predatory scientific journals and conferences, educate yourself about them to prevent falling for one of their 150,000 fake publication venues a year! This talk is a good start (DE with EN dub): media.ccc.de/v/35c3-9744-in… #phdlife #35c3

I have a dataset (or corpus) of PhD dissertations about a variety of technical topics. Looking for fun or interesting research questions to answer. #phdlife #askacademia

TL;DW: “Hold the newsreader's nose squarely, waiter, or friendly milk will countermand my trousers.”

If we link thoughts to language... this sketch by Fry and Laurie might be a nice illustration of why I answered 'infinite'.
youtu.be/3MWpHQQ-wQg?t=…

Need some external motivation to crank up your writing output? Try writeordie.com (negative reinforcement) or writtenkitten.co (positive reinforcement). I especially like the latter one since it rewards you with kitten pictures. :D #acwri #nanowrimo #phdlife

If you're looking for the insider answer to the question "What is wrong with academia?" just look at the replies in this thread. :') #phdlife #phdchat

You can start a consortium project with a group of PhDs to research the state-of-the-art in changing light bulbs, but implementation, operationalization and monetization of the research results is reserved for the industry partner.

Firas Hmida is presenting an improvement of the pipeline for lexical simplification as proposed by Shardlow (2014). Examples in French. #INLG2018 #nlproc

We can see text simplification as a monolingual translation task, and evaluate a simplification based on the relationship between the source text and the simplification. Martin et al. (INRIA and @facebookai) considered the following metrics in their experiments. #INLG2018

What kind of measurements can we use to assess the readability of text? Thomas François mentions the classical approach by Flesch (1948), i.e. base it on the number of syllables per 100 words and mean number of words per sentence, and the approaches that followed. #INLG2018

Lillian Lee's advice for PhDs: "Your goal is not to convince people that you are brilliant, but that your solution is trivial. That takes a certain strength of character. But if people think your solution is trivial, they must also think you are correct." #phdlife #INLG2018

I found a poster on formal semantics at #INLG2018 that I think you’ll like.

For future reference, here is the list of English summarization datasets from Syed’s slides. #INLG2018 #NLproc #NLP

Hilarious example by Syed et al. “When your generated output is better than ground truth”, i.e. when your summarization NLG system is better in summarizing Reddit posts than the original human author. #INLG2018

Amidei et al. conducted a survey of developments in automatic question generation, with special focus on the evaluation methods used in papers of the past five years. Interesting material, and a nice supplement to the more general survey of NLG by Gatt & Krahmer (2018). #INLG2018

Interesting: Reed et al. generated synthetic data for their experiments with neural NLG by running the inputs (attribute/value pairs from the E2E dataset) through the PERSONAGE statistical NLG system. #INLG2018

Shereen Oraby from UCSC talks about their experiments with unsupervised and supervised NN-NLG, so see whether we can control the output of neural NLG, similar to statistical generation methods. Focus: sentence scoping, distributive aggregation and discourse contrast. #INLG2018

Alnajjar & Hämäläinen open with 3 ingredients for creativity (Colton, 2008):
- skill: producing output
- imagination: creating new things (new for the creator, or for everyone)
- appreciation: ability to assess the quality of the output.

#INLG2018

Qader et al. Experimented with a coverage mechanism and pointer generator model to fix common problems of “seq2seq with attention”:
- repetitiveness
- handling out-of-vocabulary words
- “hallucinating”, i.e. generating information that was not present in the input

#INLG2018

Today's final session is about applications of automatic text generation. Speakers talk about their research on generating
descriptive product titles (@accezz et al.), messages from a virtual dietitian (Anselma & Mazzei) and company descriptions (Qader et al.). #INLG2018

Important take-away from #INLG2018: researchers need to be aware that there's a strong bias towards homogeneous output text in neural NLG methods. Models perform "better" when they are evaluated with automatic evaluation metrics, which tend to favor minimal to no text variation.

Jagfeld et al. compared word-level and character-level seq2seq models based on two different datasets, and performed automatic and manual analysis of the results. #INLG2018

More interesting posters from the first session at #INLG2018: new evaluation methods for “naturalness”, generating Japanese slogans, generating commit messages from docstrings, and flipping the message of biased headlines.

Lorenzo Gatti, another colleague from @UTwente, is explaining his work on building a multilingual NLG system for generating soccer reports. He has supplemented the NLG system with structured data about players and teams from WikiData. #INLG2018

My colleague from @UTwente Ruud de Jong is presenting his work on SimpleNLG-NL, a surface realiser for Dutch, at the first poster session of #INLG2018.

It’s nice to know that my barrage of tweets is appreciated. ;)

.@PiekVossen shows how the Pepper robot learns things about the humans and objects it interacts with. Now I wonder if, in the far future, robots will be stolen for their knowledge base, the graph of personal information they have learned from their owners. #INLG2018

There is a lot of work on information/input reliability, trust, etc. in the computer security field. I wonder if HRI uses any of this?

Slide by Piek Vossen, who is giving an invited talk on Human-Robot interaction.

Neural network irrationality, or: graphic caption generation gone wrong. #INLG2018

.@S_Gehrmann describes three neural NLG challenges:
- How can we represent our input data sequentially, for usage with seq2seq-like algorithms?
- Neural slot identification and slot filling
- Modeling latent templates that are present in the input dataset explicitly

#INLG2018

Van der Lee et al. compared statistical machine translation (MOSES), neural machine translation (OPENMNT) and retrieval methods, with and without templatization, to generate soccer match and weather reports. #INLG2018

Jan Milan Deriu talked about using a semantically conditioned LSTM to introduce more variation in generated text. The authors conducted experiments with the dataset of restaurant descriptions and related structured data from the E2E NLG Challenge. #inlg2018

This is a quote from Chris Baty’s No Plot, No Problem.

“Liberated from the constraint of constructing a pretty and proper novel, we can run free, naked and whooping, through the valleys of our imaginations.” #nanowrimo

ccnlg.org/wordpress/wp-c… via ccnlg.org

“The BLEU metric can be misleading. It is not really useful, except maybe for some instances of machine translation.” #INLG2018

Yoav Goldberg shows with an example (left) why we need to control our NLG outputs. “We forgot to specify we want different adjectives for the movie reviews.” #INLG2018

"For images, you need a bit more than strings. Luckily, the computer vision and computer images communities are doing a lot of work on transforming images to vectors." #INLG2018

"If you want to use other objects as input for neural networks, such as trees, you can easily just encode them as strings, and it will work." #INLG2018

"Don't use TensorFlow. I prefer using Pytorch or dyNet. Tensorflow is much more complicated and not really worth the effort." #INLG2018

A way of solving this is by passing the word vectors sequentially through RNN cells. There are various ways of implementing this; one example is LSTMs.

You can encode a sentence as a vector by summing over the vectors of its words; but you will lose information about word order. #INLG2018

"Neural networks are like kids, they are kind of hard to control. They just do what they like to do." #inlg2018

#INLG2018 continues with a double lecture of Yoav Goldberg about neural networks for generating text.

We have the BLEU metric for evaluating output based on document similarity and we also have the dollar metric, i.e. evaluation based on trying to sell your generated output as a work of art. #inlg2018

CC-NLG now hosts a discussion about the differences between the fields of NLG and computational creativity, and how it affects the researchers that work in either field or both fields at the same time. #inlg2018

Is that an RPG system? I'm not familiar with Microscope.

I expect that, if we can simulate Dungeon Master (GM, MC, DM) and player (PC) interactions succesfully and use this for NLG, NLG applications in the video game domain become more accessible as well. :) #inlg2018

The authors of "Powered by the apocalypse" (Tapscott et al.) propose adapting existing tabletop RPG rulesets (esp those focused on improvisational narrative) to come up with new story generators. Awesome idea! #inlg2018

In the first talk of CC-NLG after the break, Richard Doust explains a framework for suspense in narrative threads. It can be applied to the steps in a chess game to study the suspense of a particular endgame in a chess match. #inlg2018

Dear women in computer science and related fields: if you are currently pursuing a degree at a university in the EMEA region, please consider applying to the @WomenTechmakers Scholars program. The deadline is December 6, 2018. More info here: womentechmakers.com/scholars

If you want to see my designs and illustrations, follow my instagram account. In the next 60 days, I'll be posting daily. :) instagram.com/meteorietdesig…

I made a promotional poster for @NaNoWriMo based on an old US gov poster from 1937. Feel free to download it and print it. High resolution download here: jd7h.com/dump/nanowrimo… and jd7h.com/dump/nanowrimo…. Enjoy! #NaNoWriMo #nanowrimo2018

As a PhD student, I'm always trying to improve my writing skills. Reading good examples of non-fiction writing (like this article by @cduhigg) is really helpful!
newyorker.com/magazine/2018/… #phdlife

Uiteraard, maar laten we het maar niet over de financiële situatie van PhD studenten gaan hebben. ;)
Misschien tot ooit op Bessensap! o/

Bedankt voor de tip. Interessant, maar helaas niet geschikt voor PhD-funding-potjes.

De affaire Stapel ging over data-vervalsing. Dat kan in ieder vakgebied voorkomen. De ethiek-cursus was ook gemengd: het idee is dat De Wetenschappelijke Methode en ook de ethische keuzes die je als wetenschapper maakt, losstaan van je vakgebied. Dit klopt met mijn ervaring.

Ja, zeker sinds de affaire Stapel (waar ik onderhand meer uren onderwijs over heb gehad dan over andere academische vaardigheden). Heb jij nog aanraders voor wetenschappers die goed willen worden in wetenschapscommunicatie? ;)

Heel goed en zeker waar. To be fair, in de cursus was ook aandacht voor dit soort problematiek: de 'perverse prikkels van de academische wereld'.

Ik ben het helemaal met je eens! Ik zou ook liever goed onderwijs over wetenschapscommunicatie zien voor PhDs. Maar blijkbaar hebben universiteiten liever dat je als wetenschapper te voorzichtig bent, zodat de communicatieafdeling aan filteren en damage control kan doen.

In de cursus kwam er veel voorbeelden voorbij van hoe een te brede/positieve/gechargeerde invalshoek in de media een negatieve invloed op je wetenschappelijke carriere kan hebben. Het grenst in sommige gevallen aan wetenschappelijke fraude.

Een moeilijk en grijs gebied om te navigeren als wetenschapper: je wilt publiciteit, maar wel hypergenuanceerd, en dat kan niet altijd.

Eerstejaars PhDs leerden dit jaar bij de verplichte cursus Academic integrity dat je als wetenschapper verantwoordelijk bent voor hoe je je werk naar buiten draagt. Dat klopt natuurlijk, maar daarbij veel voorbeelden van hoe de media wetenschappelijk onderzoek kan verdraaien.

Yes, but not these. These books are for serial reading, since they are ordered sequels to Anne of Green Gables. ;)
I'm currently reading: Anne of Avonlea, The Company (Mickletwaith), A Discovery of Witches, Norse Mythology (Gaiman), Procedural Generation in Game Design, and more.

Uh-oh, I have just discovered the candy store for NLG and NLP researchers that is reddit.com/r/botwatch.

My Anne of Green Gables boxset arrived from Canada this morning! I am now officially ready for winter.

OH: "I work in academia! I'm allowed to write fugly code as long as it yields the intended results!"

How I get sleep deprivation in autumn: "It's late, but reading one extra chapter of Anne of Green Gables never hurt anyone..."

Me: It’s relevant for my #nlp research because it’s... a good example of text with jargon.
Supervisor: Riiiiight.
Supervisor: Very obscure jargon.
Me: ...yes.

Supervisor: What on earth is #CriticalRole and why did you use it as a dataset for your research?
Me: Well, it’s a webseries about people playing D&D and...
Supervisor: Yes, I figured that out from your results.

My research project DATA2GAME had some promotional flyers made, to explain who we are and what we do. Hopefully we can find new research participants with these. :) More info about the project on data2game.nl.

OH: "The D in DMARC stands for 'Doesn't work'."

This is mostly because of The Fault in Our Stars, since I couldn’t read more than one chapter per day without getting serious feels.

Since I’m currently reading 8 books in parallel, I’m using round-robin scheduling during my binge-reading sessions. My master’s degree in CS is clearly paying off!

Well, we can't automate mutual understanding, but we could automate facilitating better conversations between developers, for instance. Think of a github bot that asks you to clarify your coding/design choices if you open a new pull request. Like CI-tools, but for humans.

TIL the term "Open Sourceress" (via @MaritvanDijk77) #hacktoberfest

As a researcher in natural language generation (which includes dialogue generation) I beg to differ. ;) However, the success with which you can do this depends on the type, topic and quality of the conversation.

Dit is het recept: fixfeastflair.com/home/2015/2/9/…

Yes! I bought way too many graphic novels there...

I baked some kardemummabullar because I felt homesick for the Swedish cardamom buns I had in Malmö in August.

Voor alle fans van @DelpherNL: wisten jullie ook dat er een archief van @Allerhande's is waar je in kan snuffelen? albertheijnerfgoed.courant.nu

OH: this gluten-free brownie has the density of a neutron star!

@maxkreminski @annetropy @adult_witch @emshort @ptychomancer @mrj_games @xfoml You must have some useful suggestions for this list too. :)

If you're interested in procedural generation or text generation, consider adding to my list of NLG games on github. github.com/jd7h/nlg-games. I'm looking for commercial games, academic games and indie games that have *any* kind of text generation in them.

The 29th meeting of Computational Linguistics in the Netherlands (CLIN) is organized by @univgroningen on January 31, 2019. There is a Call For Papers and a Shared Task, more info here: let.rug.nl/clin29/index.p… and @29_clin. #clin29 #nlproc

Today's Natural Language Processing (NLP) lecture is about synonyms, polysemy and thesaurus-based word similarity algorithms.

@GalaxyKate The link on your homepage to Travels in Od is broken. It points to galaxykate.com/galaxy.itchio.… but I think it should be galaxykate.itch.io/travels-in-od.

Just blogged: a blogpost to accompany my cfg-based chocolate generator. judithvanstegeren.com/blog/2018/rand… #nlg

Wait but I have to finish this progress report for the funding agency... Noooooo #phdlife

Spreading the open source development love for #hacktoberfest2018 @UTwente today. Maybe I can convince some of our students to participate. The more, the merrier!

And since Twitter compressed the image, here is the link to the original: jd7h.com/dump/hacktober…

If you're looking for a #Hacktoberfest2018 poster that includes the #hackertoberfest rules, here's an A3 poster I made with the official resources from @DigitalOcean.

My Python3 chocolate flavour generator is now available in English! I'll add a flask implementation for web server deployment in the future. github.com/jd7h/random-to… #chocolate #nlg

@SamMaggs I'm a PhD student researching text generation for games. As you are the most awesomesauce game writer I know, can you recommend a book (or collection of articles) about game writing? I'm especially interested in the process itself, and evaluating the quality of output.

This article outlines two different styles of communication (focus on power vs relationships), and how differences between these styles can lead to communication problems at work. Worth a read! hbr.org/1995/09/the-po…

A thread for people who love fantasy books, language and puns.

Presenting my chocolate flavour generator at the HMI demo day @UTwente!

James Joyce or Markov Chain? This quiz reminded me of the "Ikea furniture vs metalbands" quiz... although this one is much harder. louphole.com/projets/who-wr…

Lead us in the battle against boredom and uniformity, Your Highness!

GoogleTranslate will give you a workable (ungrammatical) English version of the recipe. :)

Er staat een nieuw recept op Mosterdgeel: mega-moussaka! mosterdgeel.nl/mega-moussaka/

Thank you! :)

Thanks! :D

Thanks for all the support! ^^

I passed my PhD qualifier exam today, so I'll get an additional three years of YEAH SCIENCE! That means writing about interesting stuff, building bots, developing games, and generating game content for a living. :3

Since I started my PhD on January 1st, @year_progress is also my PhD-year progress. :) #phdlife

As someone who worked in security incident response and did on-call shifts for two years, I can't stress how important it is to be nice to your sysadmins, devops, IRs, SREs and other on-call support staff. #HugOps bit.ly/2nAHiRK

OH SNAP

TIL "word salad", a type of poem where the author just uses a collection of seemingly random words. More meanings here: en.wikipedia.org/wiki/Word_salad

We watched Hidden Figures today. :)

If you want to write grammars for generating text (for games, Twitter bots, etc.) without too much programming, check out @GalaxyKate's very user-friendly tool Tracery. Interactive tutorial here: crystalcodepalace.com/traceryTut.html #nlg #pcg

Some intermediate results. We'll probably do a second printing run, when we have found the optimal angle for printing. The parts on top and bottom are printed with less detail, and this figurine was printed lying on its back.

BTW it’s my Human Ranger named Skadi, wielding two +1 axes

Nerd dream come true: a friend has bought a 3D printer and tonight he’s printing my custom HeroForge #dnd miniature and the best part HE SET UP A PRINTER LIVESTREAM SO I CAN WATCH BECAUSE I LIVE 100KM AWAY

Love @maxkreminski's thoughts about mining games ("entitlement simulators") vs gardening games (taking care of a game environment instead of blindly using up all its resources). You can download his zine on the subject here:

How would I describe my PhD topic in normal English? "I'm a computer scientist researching text generation so that we can automatically generate game content for text-based educational digital games."

Wow, that is beautiful!

Special shout out to the organizers and attendees of the @PcgWorkshop: it was awesome. I had fun and interesting discussions with many of you. Let's keep in contact for future projects. #fdg18

To all my old and new friends in games research, I hope to see you next year and thanks for the warm welcome at my first academic conference. @FDGconf #fdg18

BTW there will be a festival about games (and other artforms) and (im)mortality in NL in november this year. theoverkill.nl/en/info/

Hilarious talk about death in video games by @EricMurnane. “I hope the player had the posh mudcrab mod, because if you are killed by a crab, it should be wearing a monocle.” #fdg18

My poster "Towards Generating Textual Game Assets from Real-World Data" for #fdg18 is now available on my website. judithvanstegeren.com/about/

@maxkreminski I really like some of the blog posts by Beck Tench. becktench.com/workflow/

That joke is so meta it took me a while to get it.

- your keys
- company server information
- phone numbers of clients
- ID cards or company badges
- readable spreadsheets
- posters featuring company security policies

Dear #CompTIAFlipMyCube contenders,

CompTIA should immediately revoke all your certificates if your submission contains the following:
- your name, address, or other PII
- company secrets
- the contents of your email inbox
- a post-it with a password
- a signed document

TIL the Finkbeiner test, a test for stories about women in science similar to the Bechdel test for gender bias in film. lastwordonnothing.com/2017/10/26/the…

Radiolab has done a podcast episode about chatbots and the blurring of the boundaries between humans and tech. wnycstudios.org/story/more-or-…

Dear author, if you mean 'random', please write 'random' and not 'aleatoric'. This kind of nonsense gives researchers a bad name.

In more scifi-men... Carth Onasi from Star Wars: Knights of the Old Republic.

Sure, if you open up your DMs.

I'm connecting from a Dutch university, so I don't think your website is blocked. The problem persists if I'm connecting from other computers (tested with multiple servers in Germany).

@valuefnd The website listed in your Twitter profile leads to a dead link/error page. Where can I find more info about VALUE?

"The most proven way to win is not to play. By abstaining from social media, you may find you have more time, a more authentic and meaningful engagement with life, and a corresponding lack of the desire to stock up on canned food and ammunition."
tools.ietf.org/html/draft-eld…

It's a Steger-saurus. ;)

Beet burger, best burger. Recipe here: greenkitchenstories.com/grilled-beet-b… #wzjve

Meet my research assistant. He is green and cuddly and a very good listener. I haven’t named him yet. #phdlife

Portal turret puppeteers at the opening of the new building 😍

Be sure to check out the text-based, procedurally generated clicker Epitaph by @maxkreminski. The best description I've seen so far is "violate the Prime Directive."

Freehold Games' @ptychomancer gave a talk about procedurally generating the histories of Caves of Qud at #GDC2018. Full talk on YouTube: youtube.com/watch?v=H0sLa1… #pcg

My favorite publication this week is Procedural Generation in Game Design (ed. @tanyaxshort and Tarn Adams) -- a book filled with thoughtful essays on #pcg in practice from all sorts of famous people. Looking forward to the section on story and plot generation! #nlg

I bought all these games from the Elegant Engineering bundle on Steam and now Train Valley has claimed my soul.

#teamblogpost

The awesome @GalaxyKate has combined tea and chatbots. #nlg

Thank you for writing long and clear README.MD's. :) Me or my colleagues from Human Media Interaction might use this project in the future.

mosterdgeel.nl/gevulde-paprik… en mosterdgeel.nl/zalmrolletjes/ en natuurlijk roast potatoes met salade. :)

And onwards! To the breach!

I'm reading Procedural Generation in Game Design (edited by T. Short and T. Adams) and I love this book. "What do we see when we stare into the abyss of Twitterbots and novel generators? And what do these twisted shadows and shapes have to tell us about ourselves?" #procjam #nlg

Well-played, Amazon.

TIL Amazon.de accepts iDeal payments, so this is the end of my bank account and the free space in my bookcase.

"I eat ALL the cheese."
Bethesda announced Skyrim: Very Special Edition at #E32018.
youtube.com/watch?v=FnEW6d…

Love these handdrawn slides by @aphyr (see @bridgetkromhout’s tweets for more examples)

It's time for career celebration donuts again! My poster about #NLG from real-world data was accepted @FDGconf! \o/

Sounds like a great job. I'm not in the US, but curious to know for my post-PhD-career: what do you look for (skills, experience, ...) in applicants for this type of job?

My fav publication this week: "Subverting historical cause & effect" by Grinblat et al., about generating coherent fictional biographies for NPCs by exploiting the human tendency for apophenia. #NLG #NLProc #PCG

"Say one more hipster thing, and I'm gonna shove you into your vintage bassoon case, ok?"

Can we please make another Jurassic Park spinoff movie but this time with Dr. Ellie Sattler the paleobotanist? I’d love to see her crashing helicopters and saving people from murderous dinosaurs WHILE identifying ferns and discussing scientific literature 😍

I guess it was just a matter of time, but Meteoriet Design is on instagram, starting today. Follow me if you want to see more of my designs and illustrations. :) instagram.com/meteorietdesign

Next week I'll be speaking about searching for scientific literature at the fifth edition of #RU-colleges at Radboud University Nijmegen.

PhD students from SIKS just got a great no-buzz-word introduction to
deep learning by prof. Eric Postma. He recommended reading ch. 1 of deeplearningbook.org and playing around with tensorflow.org to get the intuition of neural networks. #ai #machinelearning

In a lecture on data science for auditing. The challenges described are very similar to those for #infosec intrusion detection.

Finally made a start with writing a zine about my plant collection.

Yeah, I've decided not to put it in my office to spare the eardums of my colleagues.

I have bought a Filco Majestouch 2 mechanical keyboard with Cherry Blue switches and it is really I CANT HEAR YOU OVER THE SOUND OF MY TYPING

@xfoml Will you be attending @FDGconf this year?

Things NetHack taught me
* Don't get too attached to your pet
* Don't piss off your God
* Don't touch the cockatrice
* Don't eat the spiders

I read two books in the past yr that probably made me less susceptible to work-related burnout: Burn-out Dagboek by Maaike Hartjes and What About Me? by Paul Verhaeghe.

Yay! This August I'll be presenting my research (NLG for video games) at the doctoral consortium @FDGconf in Malmö! Excited to meet fellow researchers and game devs there.

Feel free to name more books or other resources. Every bit helps.

I'd love to hear your suggestions. :)

My favorite paper this week is Juke Joint by @xfoml et al., about simulating thinking NPCs in a video game and using song lyrics to influence their simulated stream of consciousness.

"Her life was heroic and meaningful, until she was petrified by a cockatrice."

At the SNT/Trimm game night learning NetHack. Apparently, there are no "winners" in this game...

If I want to read ONE book about the practice of science communication (working with press, blogging/vlogging, podcasts, writing for popular science magazines), which book do I read? Does a book like this even exist? #phdchat #scicomm

I posted a new recipe on Mosterdgeel.nl (Dutch): vegetarische bietenburgers met feta en havermout. mosterdgeel.nl/vegetarische-b…

s/tool/research paper/g

Your system needs upgradation!

I posted a new recipe on Mosterdgeel.nl (Dutch): tomaten-crèmesoep. mosterdgeel.nl/tomaten-cremes…

Prof. Mark Liberman is keeping a log of #linguistics jokes in web comics: languagelog.ldc.upenn.edu/nll/?cat=9

I think about writing zines at least once a week. I'd love to write tiny books about the mountain of unrelated things I'm interested in...

ME: phone, plz restore the notes I accidentally deleted.
PHONE: backup restored
PHONE: i deleted the past 2 yrs
ME: WAIT WHAT NO
PHONE: including all your photos
PHONE: i also restored an ancient version of the OS and a xmas themed wallpaper <3
ME: I WILL NEVER TRUST YOU AGAIN

When you regularly play #DnD over skype with friends and one of you plays a ranger

Oh wauw! Dus daar staat 'gip' voor!

My fav paper this week describes a framework for experience-driven content generation. It has a nice example of different representation levels, from direct (large search space) to abstract (smaller search space). ieeexplore.ieee.org/abstract/docum… @yannakakis @togelius

I wonder: have you watched Downton Abbey? There is a lot of drinking tea from fancy cups in that series -- I think you might like it.

If you like "procedural content generation" (creating anything with algorithms instead of by hand), check out @proc_gen for awesome examples.

As a scientist, I would love to do more science communication about my research and #phdlife. Feel free to tweet me your questions! #scicomm

Researchers, how do you use ResearchGate? It was recommended to me by the information specialists of my uni, but this is the first time I've heard about it. It looks like LinkedIn for researchers on steroids -- is it useful? #phdchat #academia

Mark and Davies did some very cool research (2013) on academic vocabulary. There is a web application based on their academic word list, which you can use to compare your writing against academic texts from a specific field: wordandphrase.info #linguistics #academia

Career achievement unlocked! Submitted my first writings as a PhD candidate to @FDGConf 2018 in Malmö. Now where are those celebratory donuts... (thanks for the idea, @lara_hogan!)

You know you have the right PhD research topic when you can cite games like Dwarf Fortress and Diablo in your papers.

The joys of #phdlife perfectly captured in a Terry Pratchett quote.

Ik kan me niet voorstellen dat het nuttig is voor jullie volgers. Het is gewoon reclame voor iets ongerelateerd aan de bieb. Jammer!

@biebenschede ik wil jullie graag volgen op Twitter, maar de wekelijkse SumAll spam-posts met jullie social media metrics schrikt me nogal af.

Allow people to take polaroid pictures with your life-sized octopus plushie? ;)

TIL WolframAlpha can give you nutritional information about your home-made dinner. Example: wolframalpha.com/input/?i=1000%…

Since I am being paid to write a dissertation, "I linguify for a living". Thank you for this wonderful piece of spiced English, @stephenfry! #linguistics #phdlife stephenfry.com/2008/11/dont-m…

"Language is the soft rain of dust that falls into a shaft of morning light as you pluck from a old bookshelf a half-forgotten book of erotic memoirs." #poetry #linguistics
youtube.com/watch?v=MSyIha…

Google for “what to do the day before your phd defence”

"If it's free, you're the product." #byebyefacebook

Check out this scientific integrity dilemma game from Erasmus University Rotterdam -- it's good practice thinking about these situations for anyone working in academia. #phdlife eur.nl/en/about-eur/c…

I’m on cake duty tonight, so I baked this cake for my fellow participants and trainers of the TSAC course Bouldering Technique I. #phdlife #poflov

Found my promotor in the Mathematics Genealogy project, and it turns out he is a descendant of Wittgenstein, Russell, Newton and Gallilei. genealogy.math.ndsu.nodak.edu/index.php

If you're living in Sweden, you must have access to all kinds of tasy pastries! :D I'm jealous.

No, but you can make them yourself with this recipe: taartenzoet.blogspot.nl/2012/10/koffie…

Making my last vacation day count with homemade Danish pastries!

If you like Harry Potter and sysadmins, go read the hilarious fanfic/tumblr The Setup Wizard about Hogwarts' first sysadmin. Poor guy. thesetupwizard.tumblr.com/tagged/setupwi…

Thank you for your talk today!

Is SmartMonster's @TriadCity still up and actively maintained? Login and game connections don't seem to work. So curious to experience the game!

Many people asked me to explain my research topic, so I made you a blogpost. I hope this helps clear things up. Questions? Let me know. :) judithvanstegeren.com/blog/2018/my-r…

Er is weer een vacature voor security specialisten bij @ncsc_nl in Den Haag. werkenvoornederland.nl/vacatures/deta… #infosec #DFIR

Nice piece! For the most part, it's also applicable from master studies -> technical jobs in general. :)

Playing around with Flask to make a small natural language generation demo.

This is so cool. matthew.wagerfield.com/flat-surface-s…

Source? Would love to read more about this. :D

"How was your day?"
"I thought about stuff. I wrote some of it down."
#phdlife

If you live in the US, you could consider getting tea from Harney & Sons. I love that brand but it’s hard to get for a reasonable price in Europe.

When Ada Lovelace was debugging her algorithm for computing Bernoulli numbers, she described the work as “so amazing a quagmire & botheration”.

And again, this is the installation part that takes the longest time to complete... wiki.archlinux.org/index.php/Inst…
xkcd.com/910/

Maker's schedule, manager's schedule. paulgraham.com/makersschedule…

Me, yesterday evening, when I was reading The Information by @JamesGleick.

Migrate from lighttpd to nginx, they said. It will be fun, they said.

I would like to buy the Moonbeams EP by Greg Barley, but I can't find the album anywhere -- except in various Spotify piano playlists. Could it be that "Greg Barley" does not actually exist? musicbusinessworldwide.com/spotify-is-cre… #fakemusic

Phew, that's quite a project. I remember it took me more than a month of dedicated reading.

That's the problem: what do you do when the book is good, but not fun to read?

I'm reading What Makes Sammy Run at the moment, and although it's an interesting book with a lot of life lessons... Most of the time, I just want to strangle Sammy.

Good books that you don't enjoy -- do you finish them, put them aside for a while or abandon them completely?

I use Jekyll, a static site generator. Less bloaty than Wordpress, but you need to be web server savvy to use it. jekyllrb.com

Thanks for the tip! I'm watching Victoria s01e01 right now and it's way more enjoyable. :D Please keep giving tv-show recommendations.

Schrödinger’s screws

I'm looking forward to meeting the #NLProc community at @CLIN28 next week. #NLP #NLG

Book Junkies Anonymous says NO DRUNK WEEDING bookriot.com/2015/11/19/how…

Some examples of words that I (computer science MSc background) had to look up: linearisation, inflection, orthography, morphology, gloss, surface linguistic form and various phrase types.

I've been reading too many #NLProc papers. My mind is stuck in a loop of "Lexicalisatie, lexicalisatie" to the tune of Shantel's Disko Partizani. youtube.com/watch?v=hc2TLJ…

...and then I suddenly became part of the @sciencebattle jury at @UTwente! Had a great time listening to the four speakers. #scicomm

Tonight I’m at @UTwente for the Science Battle theatre show. It’s a Battle of the Bands, but the bands are PhD candidates and the music is research. #scicomm

"I made an NTP joke once. The timing was perfect."

The authors of this survey paper write "We urge the reader to inspect the original paper", so go ahead and read "Children's evaluation of computer-generated punning riddles" :')

TIL there exists a research field called ‘computational humor’ #NLProc #NLG

@A_M_Nassar Just wanted to say I love this line from your #phdlife introduction. twitter.com/jd7h/status/95…

"Give some thought and time to our everyday practices; reading and writing of course, but also teaching, cooking, budgeting, walking to and from university, and falling in and out of love with your books, your stationery, and your favourite library spot." #phdlife

"Give some thought and time to our everyday practices; reading and writing of course, but also teaching, cooking, budgeting, walking to and from university, and falling in and out of love with your books, your stationery, and your favourite library spot." #phdlife

I know a few good (Dutch) #phdwomen writers. Do you only want stories from PhD students in Scotland?

Great initiative of @PhdWomenScot. 52 female PhD students blog about their experiences in #phdlife in the coming year.

I've only just realised I can probably code this myself with the Twitter streaming API and some python-twitter kung fu...

I wish Twitter had a way to filter out all tweets with more than two hashtags from my timeline and search results.

This is a strong endorsement. I will give the series a try. :)

Yes, I’m reading that too atm! The survey is well-written, but I wasn’t familiar with some of the linguistic terms the authors use to describe NLG-systems.

As a computer scientist, I was struggling to understand the linguistics content of natural language generation papers. The remedy: a few chapters of this (draft) book: web.stanford.edu/~jurafsky/slp3/ #NLG #NLProc

My twitter automation tool Zazu now supports Twitters new 280-character limit. github.com/jd7h/zazu

I'm building a Python tool to automatically search (and buy) all songs from a Spotify playlist as DRM-free MP3s from Amazon. github.com/jd7h/dandelion

Next time you plan on interrupting someone with a paper deadline, think again! blog.regehr.org/archives/1285 #academia @AcademiaObscura

"If you can't produce 140 characters of syntactically correct thought per day, maybe grad school is not for you." matt.might.net/articles/grad-…

I wish Twitter had a way to filter out all tweets with more than two hashtags from my timeline and search results.

Natural language generation can be used for building authoring aids, ie. systems that help people write routine documents, which "may considerably enhance productivity and morale". #nlg

Thanks! If you have any cool ideas for applying natural language generation to real-world security problems, let me know. I'd love to collaborate.

Today is the first day of my PhD research project. 🌞

The right way to do a New Year's Eve countdown.

Great illustration of the fact that almost every device can be hacked. wired.com/story/i-forgot…

TIL appending "+" to a bit.ly link shows you the expanded url.