As usual: finding a balance is the key. What if we got several duplicates, but every one has slightly different issue with it? Which one is a good representation of that group of probable duplicates? Other thing: what if one start with a wrong idea that as a normalization prototype we could remove all the interpunctions and than compare things (we can't) - causing huge falsely identified duplicates... So, what we can use for certain as a duplicate, and what we cant. If only a "space" would be the only problem here, but I guess it would be a good start (and I guess, various sorts of hyphens, multiple consecutive signs like !!!, etc)...

There is also always possibility to normalize for the sake of comparing sentences, but that that normalization is not carried out as a final version of the sentence. For example, always using plain space in comparisons, but if sentence use non-breaking space, leave it (as a principle to leave more complex one as a chosen example to survive deduplication process, and use simpler as normalized variant). But, it can get messy, that is: it could bring more problems. It is hard job.

I believe some people take great pain in using proper spacing character, for example non-braking space, around % in German, so forcing plain space might be wrong approach. However, maybe merging script could report sentences that differ only by a single letter? So admin can decide what to do with it.

We have some non-adopted sentences in English with the phrase:"much the best".
http://tatoeba.org/eng/sentence...rom=eng&to=und
Please, if someone would contribute some really good example when such phrase would really be used (or adopt existing one). That one indeed seem strange, at least for a non-natives, so I wonder. Thanks.

I suppose network connection was really bad, and interface didn't give a feedback that the request was already sent (the interface looked like frozen, nothing changed), the BUTTON for sending it wasn't DISABLED after the first request, and one persistent contributor more than willing to get over with it and to go to the next sentence... It happens all the time, but we usually make only one duplicate (two clicks) and than wait.

I don't think it is possible. Sphinx, search engine we use, are based on indexing words (or stems of words), not their parts.

I think I figured out what is going on. I didn't look at the code, I can't be bothered with that now, sorry. But, I have been in editing mode of one sentence (entering it), and I left that page for quite some time (half an hour, maybe) - and then return to it, and simply pressing Enter. And it was created without owner of the sentence. As if session has been lost, and request was send with invalidated session id, or something similar.
This scenario was for this one:
http://tatoeba.org/eng/sentences/show/3410366
(I had to adopt my own sentence.) But, I have noticed that previous error I was talking about, was also for the first sentence after quite some time (I guess the user had left the page opened in the browser and committed the sentence later). Hope this help. I really can't look at the code.

It's nice to be back.

CK, is it possible to make a regular Tatoeba list with those sentences, it would be easier to add translations/check if there is already one? If this turns out to be cumbersome, never mind...

It might still be some problem at the server side. For example, for this sentence
http://tatoeba.org/eng/sentences/show/2945258
eirik174 edited 5 hours ago, it didn't log user who performed the change. But, it did log correctly his changes from 11 hours ago, on this sentence
http://tatoeba.org/eng/sentences/show/2935147 .
So it is not quite isolated case, and seems to be time dependent. (like some locked table in the database preventing access to some table etc....). Probably it would need to find out when this happened for the first time, and check any change that had been performed around that time.

May I suggest that you, Silja, try to clear the cache? It seems like only few people are consistently affected. So, for me it is obvious that the problem is in your browser.

It doesn't work well if the sentence is hidden (according to the user settings) (or if the sentence is not yet shown since page is partially different cached - new logs, old sentences...). But, highlight still remains on the previous sentence (if our click on the "hidden" sentence isn't the first click). So, I would recommend that any click on log first removes previous highlight and than try to set the right one. Otherwise, it could be dangerous to provide wrong info (previous sentence as a correct information), since we might want to base our decisions on that.

I just love that new feature about clicking on the log entry. It will help a lot when we need to split mixed translations.

I think such things are very useful. It would be useful for any not so big language here at Tatoeba to have its translation in those biggest linked groups.

If you eliminate all punctuation (.,!?;:- etc.), and use 1. solution, you will get exactly one expected result.
So, try:
http://tatoeba.org/eng/sentence...rom=jpn&to=und

You may try this two:
1. ^いる?$ ( in front of it add ^ , and at the end add $ ).
2. ="いる?" ( insert your text in ="." )
The first one seems to work something, but it isn't ideal. I believe problem is the question mark here, you might get better results for regular sentence.

This wasn't the question, but this is the topic that keeps popping up, from time to time. I really don't understand is that so difficult to do the following. If you are learning the language, and it is difficult for you to came up with something meaningful, and you decide to peek somewhere for an inspiration, lets say like that. Avoid lyrics, since they are..., well, so poetic, and recognizable. But if you use some text, why simply don't change it so it would practically be totally different sentence. There are synonyms, word order that may be changed, persons changed, a lot of things.Just make it yours. That way, you will avoid getting into problems, that is, avoid putting Tatoeba into legal problems.

And to be clear. I am not proposing to take newest novel by J.K.Rowling and rewriting it down here completely. Nobody is so naive.

Use an imagination lead by an inspiration.

Ha, you are trying to play tricks with them. To lower their guard. Clever. I won't tell them, don't worry. :D

Folks! What a fierce battle is going on between Franch and Italian here! Who shall win?

A month ago, or so, tatoeba.com looked like improved version of tatoeba.org (all green, style, logo, fonts, ...). First thing I've noticed, is that it has changed login page, with google, and facebook login, centered over the page. My first thought was: look what those google's students already have done. Then, I remembered to check the domain. Luckily, I didn't log in. And I got there, in first place, by using fresh installation of browser, and typing just tatoeba (browser autocompleted with .com). Now that domain appears to be harmless, but you never know when it can vamp itself back.

Maybe we could get rid of email notifications, and get notification page for each user. Something similar as PM system here. But, it would collect all that should go to email, to a certain page. And email might only be fired if there is no logged user for a month.

Other thing about this problem that should concern all of us. There is one other website that I was suspicious about, tatoeba.com (com, not org) Is it a phishing site? Can someone confirm this is really also "ours"?