** Sentences & Translations Stats **

These stats have been updated:

http://tatoeba.j-langtools.com/transtop/

Graphs have been updated, too.

http://tatoeba.j-langtools.com/graphs.html

Thanks a lot, Bill! :-)

* Tatoeba Graphs *

New graph (Tatoeba top 10 most active languages) has been added on the page below:

http://tatoeba.j-langtools.com/graphs.html

Всегда пожалуйста. :-)

** Sentences & Translations Stats **

These stats have been updated:

http://tatoeba.j-langtools.com/transtop/

Graphs have been updated, too. Two new graphs (Top Languages Monthly Share Graphs) have been added.

http://tatoeba.j-langtools.com/graphs.html

** Sentences & Translations Stats **

These stats have been updated:

http://tatoeba.j-langtools.com/transtop/

Graphs have been updated, too:

http://tatoeba.j-langtools.com/graphs.html

Works like a charm now. Thanks, gillux!

Non-https link to the license icon (http://i.creativecommons.org/l/.../fr/88x31.png) causes security warning ("partially encrypted page") in my browser (Seamonkey). Is it possible to replace this link with the following: https://licensebuttons.net/l/by/2.0/fr/88x31.png or some other https link?

Thanks, Trang! Sad but not fatal. :-)

* Translated UI issue *

In seems that declension keys are ignored in language drop-down lists while displaying them in translated UI. For example, in Russian UI all declinable language names have two cases: nominative ('nom' key, the default) and prepositional (prep 'key'). It appears that those keys are honoured in separate strings but not in drop-down lists. That is, for example,

Sentences in {language} not translated into {translationLanguage}

is rendered in Russan as

'Предложения на {language.prep}, не переведенные на {translationLanguage}'

and is being displayed correctly while

'Search sentences in {langFrom} translated into {langTo} containing:'

rendered as

'Искать предложения на {langFrom.prep}, переведенные на {langTo} и содержащие:'

is displayed as if {langFrom.prep} was just {langFrom}.

Could this behaviour be corrected?

Unfortunately, timestamps in Tatoeba sentence database dump don't contain time zone information. I guess it's France time zone (Central European Time (UTC+1) or Central European Summer Time (UTC+2) depending on whether daylight saving time in France is in effect or not).

** Sentences & Translations Stats **

These stats have been updated.

http://tatoeba.j-langtools.com/transtop/

Graphs have been updated, too:

http://tatoeba.j-langtools.com/graphs.html

If stemming algorithm uses dictionaries, it may honour hyphens for words that can be found in dictionary and ignore them for all the rest. "Two-way" is a correct word so there are chances it exists in dictionary while "всего-лишь" doesn't and it maybe treated as two separate words. That's only an assumption, though.

Looks like a feature of the stemming algorithm that Sphinx engine uses.

Try this:

="two-way"

It depends on how accurate the result should be. The most simple algorithm is as easy as finding each word in dictionary based on full accentuated paradigm. This method is fast but it's accuracy is limited by dictionary size. In addition, it can't deal with homographs. More accurate approach involves deep morphological and semantic analysis. Unfortunately, there are no usable Russian analyzers in public domain that I'm aware of.

Unlike my kind colleagues I don't always use ё and I can't promise I will. That's not because I'm insensitive to the problems of learners. I just want to write sentences the way I do it in real life focusing on more important aspects like naturalness and accuracy.

I think it would be more practical and useful for Russian language learners if our great programmers implemented editable alternate script feature for Russian that would allow us to provide versions of original sentences with all ё's and stress marks on every word.

** Sentences & Translations Stats and Graphs **

These stats have been updated.

http://tatoeba.j-langtools.com/transtop/

Graphs page has been updated, too. Additionally, two new graphs (user activity distribution by weekdays and hours) were introduced.

http://tatoeba.j-langtools.com/graphs.html

Really nice recordings. Thanks, huizi99-san! We want more! :-)

** Sentences & Translations Stats **

These stats have been updated.

http://tatoeba.j-langtools.com/transtop/

Various graphs have been updated, too:

http://tatoeba.j-langtools.com/graphs.html