Pages in topic:   [1 2] >
Leveraging repetition hits
Thread poster: Omer Shani
Omer Shani
Israel
Local time: 10:28
Member (2012)
Apr 25, 2016

Hi collegaues,

We have been using memoQ ever since our establishment in 2012, and we are really satisfied with it.

We are now facing a more challenging client, which means we need to push memoQ to the upper limit.

The main issues are about repetition hits, specifically for this client.

The client may have the following segments:

* You must pay 360$

* You must pay 450$

Since they are about cyber training, they also have deliberate typos along with the correct ones:

* Your email quota axceeds 200GB
* Your email quota exceeds 200GB

As you can see, there are very minor changes between segments, and still, memoQ does not recognize it as repetitions. which makes the client being frustrated, for they were hoping to save some from these breakdowns.

Is there a solution to leverage the repetition algorithm, in order to tell memoQ to ignore numbers or special incidents, just to increase the repetition hits? Or any other solution?

Thanks you

Omer Shani
BlueLion Language Services


Direct link Reply with quote
 
Stanislav Okhvat
Local time: 11:28
English to Russian
Correct typos prior to import Apr 25, 2016

Hello Omer,

No, you can't override memoQ's matching algorithm. You can automatically insert the best match there is, but you will need to review each automatically inserted translation because you can't be sure whether the low percentage is caused by a typo or a different word.

I would suggest that you spellcheck the source text using Microsoft Word (if this is Word format) and correct each deliberate typo, but you will need to do this for every revision of the document. If the amount of such deliberate mistakes is large, this may turn out to be more economical than to review each partial match.

Best regards,
Stanislav


Direct link Reply with quote
 
Omer Shani
Israel
Local time: 10:28
Member (2012)
TOPIC STARTER
Clearing the required topic Apr 25, 2016

Hi Stanislav,

I'm afraid my intention wasn't clear enough.

The typos in the source segments are deliberate ones, since the client wants to demonstrate to their clients how to avoid cyber attacks (which often can be identified by typos).

But since these segments are very similar to other correct segments, and often other segments are differentiated by numbers only, (as with the 360/450$ example), my question was how to increasing repetition leveraging, if possible any how.

Is there a way to set up an analysis setting, specifically to this client, and advise memoQ to detect almost-identical, and ignore specific characters, to be identified as repetitions?

Omer Shani


Direct link Reply with quote
 

Thomas T. Frost  Identity Verified
Member (2014)
French to Danish
+ ...
Still not clear Apr 25, 2016

If the mistakes are deliberate, then I suppose you also want deliberate mistakes in the translations. In that case, it requires a manual 'mistranslation' by the translator, so why would you expect such mistakes counted as repetitions?

The translator also needs to verify that the numbers are correct, I suppose. He or she can't just click to approve the automatic translations without checking them.

At some point, you need to ask yourself if you want a competent translator or just the cheapest one, regardless of quality. The more you slice off the rates, the more the good translators will decline.


Direct link Reply with quote
 

Lincoln Hui  Identity Verified
Hong Kong
Local time: 16:28
Member
Chinese to English
+ ...
Variables Apr 25, 2016

For segments where only the number gets changed, you can choose to simply have a string that stays "You must pay %". How to replace % with the number of your choice is your own decision.

As for the deliberate typos, you better damn well pay in full for those.


Direct link Reply with quote
 
Omer Shani
Israel
Local time: 10:28
Member (2012)
TOPIC STARTER
Clearing the issue Apr 25, 2016

I agree with you Thomas, but in a business state of mind, one can ask himself why paying twice for almost identical segments.

If one segments goes:

* You must pay 360$

And the following one is:

* You must pay 450$

Why doesn't it counted as repetition? or better one: How do I train my memQ to ignore the number issue and detect the text alone?

And if we have typo issue ( which should be reflected in the translation obviously, when the only difference is the typo itself, as in the following example:

* Your email quota axceeds 200GB
* Your email quota exceeds 200GB

I'm not interested in lowest bidder client, and the client himself is interested in continue working with us, though we're not a low-fare translation agency, but I'm looking for some out-of-box memoQ solutions, to keep this client more satisfied.

Omer Shani


Direct link Reply with quote
 

jyuan_us  Identity Verified
United States
Local time: 03:28
Member (2005)
English to Chinese
+ ...
So your client wanted those similar segments to be considered as repetitions? Apr 25, 2016

That's too much.

Direct link Reply with quote
 

Thomas T. Frost  Identity Verified
Member (2014)
French to Danish
+ ...
Not a repetition Apr 25, 2016

If the translator needs to intervene, it isn't a repetition. You can count it as a repetition when the translator has nothing to do, except just approving the automatic insertion of the repetition.

But that won't work here.

MemoQ will rate it as a high-percentage match, so you can count a percentage of the full word rate for that segment but neither 100 % not 0 %. That's quite common.


Direct link Reply with quote
 
Omer Shani
Israel
Local time: 10:28
Member (2012)
TOPIC STARTER
Trados vs. memoQ Apr 25, 2016

Hi,

After Googling a little bit, I've found a similar post, when the translator showed the following examples, from memoQ & Trados: (The following is a quote from that discussion)

"Source:
My father has 5 horses.
My father has 6 horses.

memoQ analysis (no homogeneity):

10 No match
0 Repetitions

Trados analysis (No internal fuzzy match leverage):
5 No match
5 Repetitions"

How can that difference be from Trados & memoQ?

How can Trados identify the segment as reps, while memoQ doesn't?

I truly believe that something is not that perfect here, that we can still leverage our work and business.

If the only work that a translator has to do is change a number, without harming the syntax and linguistic correctness, then why shouldn't memoQ offer that possibility, especially if a market leader as Trados does offer that possibility, somehow?

Thank you

Omer Shani


Direct link Reply with quote
 

Thomas T. Frost  Identity Verified
Member (2014)
French to Danish
+ ...
Plenty of ways to do it Apr 25, 2016

MemoQ will give you the following match percentages:

95%-99%
85%-94%
75%-84%
50%-74%

in between repetitions/100 % and no match.

Just use them, and define a percentage word rate for each type of match.


Direct link Reply with quote
 
Omer Shani
Israel
Local time: 10:28
Member (2012)
TOPIC STARTER
Not fizzues, reps Apr 25, 2016

I'm not talking about fuzzies, I'm talking about repetitions, within a specific file.

Is there a way, any way, to define the analysis settings of memoQ, as we can define QA settings and other features?

Omer Shani


Direct link Reply with quote
 

Thomas T. Frost  Identity Verified
Member (2014)
French to Danish
+ ...
Going around in a circle Apr 25, 2016

I think we're going around in a circle now. They are not repetitions but you want them to be.

If you consider there is no work involved for the translator, then you could make the number changes and the intentional mistakes yourself so the translator will only get the pure and true repetitions to deal with.

If you want work done, you need to pay for it. If you don't want to pay for it, you can do it yourself and keep the translator out of the number changes and intentional mistakes.

You don't want to use fuzzy matches either, and you don't even say why. This sounds rather hopeless to me.


Direct link Reply with quote
 
Pete in Finland
Finland
Local time: 10:28
English to Finnish
+ ...
Because Apr 25, 2016

Omer Shani wrote:

If one segments goes:

* You must pay 360$

And the following one is:

* You must pay 450$

Why doesn't it counted as repetition?


Because it is not a repetition.


Direct link Reply with quote
 
Omer Shani
Israel
Local time: 10:28
Member (2012)
TOPIC STARTER
But Pete... Apr 25, 2016

Pete in Finland wrote:

Omer Shani wrote:

If one segments goes:

* You must pay 360$

And the following one is:

* You must pay 450$

Why doesn't it counted as repetition?


Because it is not a repetition.


The only difference between these 2 segments are the numbers. I can agree that these segments are not repetitions if they had a a word which differs. But the only difference is a number, which has low or even none, impact on the syntax and linguistic structure.
Once you have translated the 1st segment, I'd expect memoQ to know that the following one is almost identical. That is why I'm asking about the analysis engine, and refine it a little more.
As I noted earlier in the Trados example, if Trados can identify it, why memoQ can't? and if memoQ can, how to set it up?


Direct link Reply with quote
 

Michael Joseph Wdowiak Beijer  Identity Verified
United Kingdom
Local time: 08:28
Member (2009)
Dutch to English
+ ...
try selecting "TRADOS 2007-like" Apr 25, 2016

Omer Shani wrote:

Pete in Finland wrote:

Omer Shani wrote:

If one segments goes:

* You must pay 360$

And the following one is:

* You must pay 450$

Why doesn't it counted as repetition?


Because it is not a repetition.


The only difference between these 2 segments are the numbers. I can agree that these segments are not repetitions if they had a a word which differs. But the only difference is a number, which has low or even none, impact on the syntax and linguistic structure.
Once you have translated the 1st segment, I'd expect memoQ to know that the following one is almost identical. That is why I'm asking about the analysis engine, and refine it a little more.
As I noted earlier in the Trados example, if Trados can identify it, why memoQ can't? and if memoQ can, how to set it up?


When running Statistics, try selecting "TRADOS 2007-like", instead of "memoQ". I'm not sure, but I think this changes the way numbers are counted.

MJWB


Direct link Reply with quote
 
Pages in topic:   [1 2] >


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Leveraging repetition hits

Advanced search






TM-Town
Manage your TMs and Terms ... and boost your translation business

Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.

More info »
Déjà Vu X3
Try it, Love it

Find out why Déjà Vu is today the most flexible, customizable and user-friendly tool on the market. See the brand new features in action: *Completely redesigned user interface *Live Preview *Inline spell checking *Inline

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search