MS Translation SOAP Interface Has Transaction Limits?

Question

I am using the Translator SOAP interface and it is working quite well....right up until it quits with the error message "The HTTP service located at http://api.microsofttranslator.com/V1/Soap.svc is too busy."

It seems to always error out like this after some number (100+ ?) of translations have successfully completed.

Am I doing something incorrectly, does the service have volume/transaction count limits, or.....?

Any suggestions (please and thank you)?  Great service just before it quits!

Answer 1

Hi L,

you are hitting an intentional throughput limit, which is there to protect the MT service, and other users of the service.
You can do one or both of these things:

- Whenever you hit the limit, sleep for two minutes and then retry.
- Stop new requests when you have hit 50 in a minute.

In practice this 50 number may be a bit lower, try how it works with your content.

If you need a higher throughput limit for commercial use, let us know at [e-mail].

Chris Wendt
Microsoft Translator

Answer 2

Chris,

Thank you for the reply.  I understand the message and the reasons for the limits.  I will be sending an email to the referenced address before long so we can begin to prepare for production implementation which is a few weeks off.  Until then, I think we can work around this during development.  I had a feeling something like this would be the answer.

Thanks again!

Answer 3

I had the same problem and I tried merging the pieces of text for translation into one, spliting them with a separation string.
However I now get another exception:

The formatter threw an exception while trying to deserialize the message: Error in deserializing body of request message for operation 'Translate'. The
maximum string content length quota (8192) has been exceeded while reading XML data. This quota may be increased by changing the
MaxStringContentLength property on the XmlDictionaryReaderQuotas object used when creating the XML reader. Line 76, position 205.

Can't you make the quota a little bigger ?
I quess I have to combine your technique with mine to get what I want...

Also your tanslator seems to change the formatting of the input text.
E.g. I translated the contents of an SRT (subtitle) file here and the translator removed newlines, added spaces, changed the case of untranslated text, etc.
I think that all of these are bad features (not to mention that they make my method -described above- become problematic :( ).
Google's tranlation tool seems to preserve the formatting better. Perhaps you could add an option whether the translator should reformat the translated text or not.

Answer 4

Hi Kostas,

the limit protects the service and its users. But indeed you may go up to the size limit and you'll be fine. When combining segments and then inserting splits and markers, be sure to never cut sentences into two pieces. The engine needs to see a sentence fully intact to provide a reasonable translation, and a sentence cannot stretch over multiple requests.
If you combine sentences that do not end with a language-appropriate sentence end punctuation, be sure to add one (for instance a period followed by space in English).

Chris Wendt
Microsoft Translator

Answer 5

Yes, I do whatever I can to give the sentences right.
But the changes in the formatting I mentioned above (I have edited my previous post) introduce some trouble (e.g. the translation adds spaces inside my separator strings, changes the case of non-translated pieces of text -- which usually are my separator stings -- , etc).
Anyway... I'l find a way to solve my problems.

Thanks,
Kostas.

Answer 6

I think I've found a nice separator.
I separate sentenses using the character '¶' (U+00B6).
It seems to work quite fine most of the times...

You know, it would be nice if you treated a special symbol (e.g. '§'  U+00A7) like a fullstop and leave it untranslated/unchanged in the output.
In this way, a programmer could translate many sentenses at once (using '§' as a separator) and without having to add fullstops that don't exist in the original text (e.g when translating movie subtitles).
Can you do sth like that please ?

Your translator fails to translate the words below (it leaves them as is):
οδηγήσω
προσλαμβάνεις
Κουνηθείτε
καρναβάλια
κατέβηκες
έφθασες
υπόνομο
σκάσε
ξεκουραστείς
έγδαραν
μάγκες
Θεούλη
σοβαρολογείς
ακούσεις
σκοτωθώ
Κανείς
πεθάναμε
αστοχήσεις
Ξεκλείδωσα
Κρατηθείτε
Κρατήσου
εξοντώσατε
απολύσω
............

Do you ever see the translations with low rating (to improve your translator) ?

Your translator is a great tool !
I hope you improve it !

Answer 7

Hi Kostas,

the ¶ character is just a normal character. It is not a sentence breaking character. It will worsen the translation when interpreted as part of the sentence.
I do think you have a good point in suggesting a universal sentence breaking character.

Chris Wendt
Microsoft Translator

Answer 8

Kostas,

the API is HTML aware, so you can use HTML sentence-breaking tags, for instance <p>. In fact any tag that is not known in HTML to be internal (non-sentence breaking) is an external (sentence breaking) tag.

The non-sentence breaking tags are

a
a0
a1
a10
a11
a12
a13
a14
a15
a16
a17
a18
a19
a2
a20
a3
a4
a5
a6
a7
a8
a9
acronym
b
b*
bdo
big
div
em
font
g0
g1
g10
g11
g12
g13
g14
g15
g16
g17
g18
g19
g2
g20
g3
g4
g5
g6
g7
g8
g9
i
s
small
span
strong
sub
sup
u

where b* means any tag starting with the letter b. Any other tag breaks the sentence.

Answer 9

Thanks a lot.

Wrapping each sentence in a "<p>...</p>" pair solves the problem.

Answer 10

Good, thanks for letting us know.
Also thanks for reporting the untranslated words.

Chris Wendt
Microsoft Translator

Answer 11

Hi L,

you are hitting an intentional throughput limit, which is there to protect the MT service, and other users of the service.
You can do one or both of these things:

- Whenever you hit the limit, sleep for two minutes and then retry.
- Stop new requests when you have hit 50 in a minute.

In practice this 50 number may be a bit lower, try how it works with your content.

If you need a higher throughput limit for commercial use, let us know at [e-mail] .

Chris Wendt
Microsoft Translator

Chris, what do you mean by "In practice this 50 number may be a bit lower?".

In my opinion, it makes a lot of sense to have a specific limit and not a "may be a bit lower"...

Your support seems very good and we are considering using your API, but can't we get a more precise number of the amount of requests we can make for a certain period of time?

It's a little sacary to not know if we are reaching the limit or not. And this is specially important for what we are trying to achieve.

Please reassure us. Thank you

Answer 12

Correcting myself: if you stay below 50 requests per minute and below 1000 characters per request you are quite safe. We'll fail the request only under extreme situations.