12,382 karma · joined May 31, 2012
Over time I ended up just using Workflowy for a grocery list and it was relatively slow for that - nothing drastic, just a few seconds to load when I fired it up - so I changed to something simpler for that specific task.
https://pubmed.ncbi.nlm.nih.gov/29956364/
> The cookie groups were provided with 500 g of cookies at the beginning of the session; this corresponded to an over‐supply of cookies to avoid limiting students to taking one cookie only. The teachers were instructed to comment on the availability of the cookies only by saying: ‘I've brought some chocolate biscuits.’ Teachers were allowed to eat one chocolate cookie themselves to break any reservations about taking the first cookie. The control groups received no chocolate cookies.
> If students brought cookies or other sweets for their group, the teachers were instructed to report this to the study group.
> At the end of the session, all students were asked to fill in the anonymous end‐of‐session evaluation form. While the students answered the questions, the teachers were instructed to leave the room. The forms were collected in a ballot box. After all students had left the room, cookie consumption was determined by subtracting the weight of the remaining cookies from 500 g.
The full article also seems to indicate cookies had no effect on test scores; see this Twitter thread, which is where I found this link.
You only note negative things, not positive things. This colors the whole post.
You use italics and "Oh,", which makes it more conversational, and it sounds exasperated to me.
Your questions at the end could be construed as rhetorical, rather than constructive, partly because you suppose an answer.
Suggesting that the games that are there exist because they are "easy to program" could be seen as a negative comment on the creator's programming ability or drive.
As the poster of this link and someone who enjoys the games a lot, I think your criticism about lack of variety is valid (if out of place for a collection of free games) and your phrasing wasn't aggressive or anything, but the above are some points that I watch out for in my own writing, particularly the first one - even if I don't like something, I find it useful to point out positive aspects to make it clear I'm trying to be constructive.
Fleisch-Kincaid is for written documents. It's dubious whether it's really of value even in that limited circumstance, but it makes no sense for transcriptions of written speech, where punctuation is basically up to whoever's transcribing it. The metric is strongly influenced by sentence length, so you can see why it would cause problems.
Language Log has written about this and why it's ridiculous in detail before.
UniDic has a rule that a "word" (or a "short unit word" 短単位, the unit used in UniDic) can't be more than two "minimal units", where an example of a minimal unit is a kanji with a Chinese reading. So 麩菓子 can't be a word, and in this case it's 菓子 with 麩 on the front.
You can read more about short unit words here.
https://unidic.ninjal.ac.jp/glossary#suw
It's not a perfect tradeoff, and I think it can be improved, but it does have its advantages.
https://github.com/polm/fugashi/issues/23
(Also, if you learned about Chrome's Japaense tokenization on HN, that was me too!)
There's a Japanese site for tech blogging called Qiita that puts a warning at the top of any post that's over a year old. While a scorched earth policy could be more satisfying (and potentially safer), using a warning might make it easier to get buy-in.
Here's a random Qiita post with a warning (yellow box):
http://www.immi-moj.go.jp/newimmiact_1/en/point_3-4.html
To add to what bojo said, when you apply for it at the airport these days you basically just fill out a tiny piece of paper and they say "OK, see you when you get back" if you're going away for less than a year. I don't go overseas much but if memory serves the process is basically the same for PR and non-PR visas.
I don't really know much about the implications of only coming back once every five years or so.
At my first job I was a "contract employee", which was a much less stable position than I realized when I signed up, and I couldn't get more than a one-year visa either. Later I changed jobs and became a full employee and got a 5-year visa without issue.
I was considering that route to self-employment before I learned about the change in the HSFP visa.
> (I'd even be willing to pay taxes in Japan!)
If you are a resident of a country it's only reasonable to expect you to pay taxes there...
https://www.jetro.go.jp/en/invest/setting_up/section3/page7....
It looks like in some documents the term "permanent resident" is used to refer to "residents of more than five years" or similar, but it's not a reference to visa status.
https://www.japantimes.co.jp/community/2016/10/30/how-tos/lo...
Besides the terminology issue, I suspect that part of the confusion may be that the PR application does require proof you've been paying taxes, and Immigration can ask for further proof. In my own case my annual tax statements were enough.
Remote: Yes (only)
Willing to relocate: No
Technologies: NLP, spaCy, pytorch, Python
Email: howdy@cotonoha.io
Github: https://github.com/polm
Need help with an NLP project? I've implemented several systems from scratch, working with Japanese and English over the years, and can help you set something up or improve your data pipeline.
My one-line open source resume: I maintain the most popular Japanese tokenizer in Python, and added most of the Japanese support in spaCy, and have recently fixed issues with Japanese in Transformers and sacrebleu.
Outside open source I worked in e-commerce for several years and have been supporting work on a fashion search app as a continuation of that.
https://web.archive.org/web/20160304171428/http://www.pixen-...
https://kylecourtney.com/2020/03/11/covid-19-copyright-libra...
If you are unfamiliar with Twitter, you could consider a user's timeline like their "home page", since they control it, and an RT could be considered a form of "reposting", because you put it in places it had not been before.
I don't think that's a reasonable line of thought, but it is consistent.
https://www3.nhk.or.jp/news/html/20200721/k10012526381000.ht...
The photographer wants to email the retweeters. The reason isn't stated but he could send them a bill for use of his image and threaten to take them to court if they don't pay it.
This is not something a reasonable person would do, but it is technically within his rights under the law. There are occasionally people online in Japan who send bills to people (mainly magazines) who quote them or use images on Twitter in ways they don't like.
Japan does not have fair use, instead it has a number of specific exceptions to copyright. For example, you can take photos of things semi-permanently installed in an outdoor area and use those photos for non-commercial purposes, even if you photographed a 2D artwork that might have copyright. You can read the copyright law in English here:
https://www.cric.or.jp/english/clj/cl2.html#art32
The most generous exception in the law (except maybe the search engine one) is for "quotation", Article 32. For quotation you must use only as much of the work as is necessary for commentary, clearly separate it from your own content, clearly attribute it, and the quoted content must be subordinate to your own content. Regarding the last point, that means you can't just say "this is a nice painting" and post the image, you have to provide some commentary on it. (No, that's not clearly defined, so you'd need to work it out in court if there was a dispute.)
In practice this doesn't matter much on social media, and things work out much like places where you have fair use, but sometimes people try to enforce all their rights. Usually the people they are upset at take down their posts, but I guess in this case that didn't happen.
One really unfortunate side effect of this is that Japanese Wikpedia has very few images compared to other Wikipedias. Things that would be fine in an article, like book covers, are excluded because their Wikimedia file page would not have commentary and, so the thinking goes, would be infringement. (I remember reading the discussion that came to this conclusion on a Wikipedia discussion page years ago, but haven't been able to find it since.) This is why the Japanese article for Anpanman has a photo of a mural, for example, while the English article has a book cover.
For moral rights (著作人格権), these actually also exist in many places besides the United States. Practically speaking, if you license a work for use it's pretty common to have a clause saying you will not exercise these rights, and if you're in a mutually beneficial relationship any concerns over how you're credited can usually be negotiated. In this particular case the use was completely unauthorized so that didn't happen. (I am still confused about why the poster didn't just remove the image when asked.)
Usually people with their rights "infringed" by unauthorized posting on Twitter don't care. If they do care, the poster usually takes the image down. I guess that didn't happen in this case, which is why it went to court, but it rarely gets that far.
Note that usually people are reasonable about it, and that it's also extremely common to have contracts state the right will not be exercised. This case is unusual because the use of the photo is completely non-consensual.
https://arxiv.org/abs/1703.00854
I can't say I followed all the proofs, but it seems that under certain limited assumptions about labelling functions they prove their generative function can do well.
Reading Snorkel it initially sounded like magic in the bad way, but this does make it clear that if your labelling functions are garbage or have certain kinds of problems there's nothing they can do about it.
Even leaving aside the generative model I think the focus on function-based data bootstrapping is great, which is why I've been following Snorkel's projects for a while.