CafeTran

Dear Michael,

An excellent occasion to present this great new feature to the translation community! Good luck!

M

IK: and punctuation characters should be omitted when you create a pattern now

I'd think they're part of a segment, and can even change the meaning - and therefore the translation - of it. And I suppose you mean "end" punctuation, surely not all punctuation?

H.

clickable links:

http://www.proz.com/virtual-conferences/668/program/10534 (Leveraging voice recognition technology for greater efficiency panel)

http://www.proz.com/virtual-conferences/668/program/10528 (Glossary & Terminology)

Cool!

By the way, I am going to be one of the panellists in the upcoming ProZ.com (2015 virtual conference for International Translation Day) webinar titled ‘Glossary & Terminology’, and I will try to say a few words about this new feature!

See: http://www.proz.com/virtual-conferences/668/program/10528 (Jeff hasn't put me on the page yet)

I am also going to take part in the one on speech recognition (‘ Leveraging voice recognition technology for greater efficiency panel’), where I will be discussing ways to use Dragon and SR in general when translating, and in particular in CT.

See: http://www.proz.com/virtual-conferences/668/program/10534

If anyone here has any stuff they'd like to ask about, or for me to talk about, let me know!

Michael

I have started a new topic under the Tips and Tricks category to share my petty know-how about the segment pattern matching feature.

Good day!

 

Dear Igor,

I want to have your clarifications on several points about segment pattern specifications.

In glossaries,

1. Synonyms do not work on either side.

Example:

use by {1} of {2};use of {2} by {1}

In both glossaries and memories,

2. Possessive case does not work.

Example:

{1}'s use of {2}

3. Hyphenated compound does not work

Example:

The {1}-{2} relationship is ...  (e.g., The employer-employee relationship ...)

Cheers,

Masato

@Igor: If I have my glossary set to ‘Regular expressions only’, and I want to store Segment patterns in it, do I need to proceed them with the ‘?’, or will this be interpreted as a regular expression and mess things up?

According to my test so far, when a segment pattern is This is {1} (without a period) and the source sentence is This is it. (with a period at the end), that pattern did not work. 

It also does not work for This is "it" (quoted) unless "it" is in a resource.

A question to Igor.

The length of an exact fragment that may be auto-filled is not limited, as I mentioned earlier, right?

Peace,

re: P.S. "Period" at the end cannot be missed. It seems "Do not match" (in Options > Memory) does not work for segment patterns.

maybe replace the period with: .?

(any character, once, or not at all)

A discovery about this feature after an hour of trial.

A variable may or may not be one word. It can be filled by an exact fragment or glossary entry of any length.

So, when you have, a segment pattern "{1} will increase the temperature of {2}."

This works for a sentence like:

Placing this device in a hot place will increase the temperature of the device.

However, you must have both "placing this device in a hot place" and "the device," either as a separate glossary entry or an exact fragment.

Cheers

P.S. "Period" at the end cannot be missed. It seems "Do not match" (in Options > Memory) does not work for segment patterns.

yet another attempt at a name:

Subsegment variables

>how does: All the leaves are {1} and the sky is {2}. [TAB] El cielo es {2} y todas las hojas son color {1}.

>relate to what I already do:

>All the leaves are .+? and the sky is \w+. [TAB] El cielo es XXX y todas las hojas son color XXX.

Segment variables (how about that for a name?) are more flexible: you can change the order of them (compared to the SL). They are also easier to insert, especially with a dedicated button. Hence, support my REQ by liking it.

I am working on an article for the wiki.

PS: I'm not entirely convinced that ‘Segment patterns’ is the best term for this.

I understand that calling it ‘Sentence patterns’ might not be optimal either, because people might use it for pieces of text that are shorter or different than sentences. However, I think the term ‘segment’ is a bit abstract and vague. What's important about this feature is the patterns that it allows you to work with, rather than the fact that it operates on individual segments in CafeTran. Having given it a bit more thought now, I'd actually prefer just calling it ‘Patterns’, on its own. Or even ‘Cloze patterns’, as suggested in the recent Proz thread.