[brailleblaster] Re: LiblouisUTDML and backtranslate string
- From: Keith Creasy <kcreasy@xxxxxxx>
- To: "brailleblaster@xxxxxxxxxxxxx" <brailleblaster@xxxxxxxxxxxxx>
- Date: Mon, 27 Jan 2014 12:38:00 +0000
Yes, and also the semantic action files I've looked at that were generated have
all kinds of references in it that are not valid for anything other than the
specific document they were generated from and in some cases have references to
elements (tags) that are not even part of the document type they were intended
to prototype. Doing a proper job of cleaning up these files is about as much
work as just taking a DTD or schema and creating the semantic action file from
scratch. A little like cleaning up a bad OCR document.
-----Original Message-----
From: brailleblaster-bounce@xxxxxxxxxxxxx
[mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of Michael Whapples
Sent: Monday, January 27, 2014 7:06 AM
To: brailleblaster@xxxxxxxxxxxxx
Subject: [brailleblaster] Re: LiblouisUTDML and backtranslate string
My understanding of the difference between the document specification and
something based on document instances is as follows:
* A document instance might not cover all parts of the document specification,
thus basing semantic action files on instances may lead to incomplete support
of a document format. As an example, do all documents you write contain all
levels of heading supported by a format?
Speaking for myself I know I rarely, if at all make use of heading level
6 in my documents. Other things also may be missed out.
* Sometimes a document instance might have non-standard extensions added by
software. Basing a semantic action file on such a document instance will mean
that the semantic action file deviates from the document specification by
supporting non-standard extensions. While in the best case this simply may mean
liblouisutdml will process these additional extensions with no impact on a
specification compliant document, this may not always be the case particularly
should a future version of the document specification be incompatible with
these non-standard extensions then liblouisutdml will no longer comply with the
standard document specification and may perform incorrectly.
Michael Whapples
On 26/01/2014 13:34, Larry Skutchan wrote:
> You make some great points. It is one of those things that a blind user
> naturally wants to do, but perhaps a reading system app would be more
> appropriate for that. It seems like opening a BRF has about as much value as
> opening a plain text file. Braille Blaster is designed to work with
> structured content, and neither of those provides much in that area. There
> are plenty of other tools out there to convert simple files to include a bit
> more structure, so I agree we should be focusing resources on meeting the
> unique requirements of this project.
> Can you elaborate on the following:
> What about bringing liblouisutdml semantic action files into a state
> to match the format specification rather than document instances (
>
> -----Original Message-----
> From: brailleblaster-bounce@xxxxxxxxxxxxx
> [mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of Michael
> Whapples
> Sent: Friday, January 24, 2014 3:37 PM
> To: brailleblaster@xxxxxxxxxxxxx
> Subject: [brailleblaster] Re: LiblouisUTDML and backtranslate string
>
> My view is that opening brf is low priority and probably a distraction from
> more important features at the moment.
>
> My understanding of what BrailleBlaster is, if summing it up in one
> sentence: A compelling tool for creating Braille/tactile versions of
> documents.
>
> As brf is already Braille there is not such a need to import one if the
> output is going to be Braille.
>
> If including open brf is to get an epub/nimas/other ebook format then how
> does it contribute to the main goal? It might be worth while work but is it
> really work for another project? Is inclusion in BrailleBlaster really just
> the icing on the cake?
>
> Also considering the difficulties of back translation, which Susan described
> well, is this really time well spent at the moment?
>
> What may be more important: Is there a math/chemical formula/music/etc
> editor, would it be more valuable to support all parts of a document?
> Are there other computer formats (eg. word document, PDF, ODT, etc) which we
> want to support? What about bringing liblouisutdml semantic action files into
> a state to match the format specification rather than document instances
> (this comment has been made by others a few times in the past). Also what
> about locked Braille? Better documentation.
>
> Michael Whapples
> On 24/01/2014 19:00, Keith Creasy wrote:
>> Michael.
>>
>> I'm listening. Are you saying that you don't see a need for a user to open a
>> BRF into BrailleBlaster to work on it? If everyone elase agrees then maybe
>> we don't need to do it. That does happen at APH but I suppose our
>> transcribers could just use their existing tools for that. If others don't
>> see it as important then I'm fine with dropping it.
>>
>>
>> -----Original Message-----
>> From: brailleblaster-bounce@xxxxxxxxxxxxx
>> [mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of Michael
>> Whapples
>> Sent: Friday, January 24, 2014 1:56 PM
>> To: brailleblaster@xxxxxxxxxxxxx
>> Subject: [brailleblaster] Re: LiblouisUTDML and backtranslate string
>>
>> May be my question should have been: Is this a scenario which BrailleBlaster
>> is meant to support?
>>
>> I infer from your answer that this might be a yes although I am not sure
>> really why it should be.
>>
>> Michael Whapples
>> On 24/01/2014 18:44, Keith Creasy wrote:
>>> I don't think so but it will be soon.
>>>
>>> -----Original Message-----
>>> From: brailleblaster-bounce@xxxxxxxxxxxxx
>>> [mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of Michael
>>> Whapples
>>> Sent: Friday, January 24, 2014 8:36 AM
>>> To: brailleblaster@xxxxxxxxxxxxx
>>> Subject: [brailleblaster] Re: LiblouisUTDML and backtranslate string
>>>
>>> Is this a scenario supported by BrailleBlaster?
>>>
>>> Michael Whapples
>>> On 24/01/2014 13:25, John J. Boyer wrote:
>>>> Back-translation would also be used if the user wanted to open a
>>>> brf file.
>>>>
>>>> John
>>>> On Fri, Jan 24, 2014 at 06:25:31AM -0500, Keith Creasy wrote:
>>>>> Hi all
>>>>>
>>>>> Back-translation only applies to BrailleBlaster in reference to six-key
>>>>> Braille input and the desire to have both a text and Braille view.If that
>>>>> is for some reason not the case then I guess we need to decide that.
>>>>>
>>>>> Thanks.
>>>>> Keith
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Sent from my iPad
>>>>>
>>>>>> On Jan 24, 2014, at 5:23 AM, Larry Skutchan <lskutchan@xxxxxxx> wrote:
>>>>>>
>>>>>> Perhaps APH expanded the project's scope by targeting the development of
>>>>>> a tool that transcribers could use to produce effective, accurate
>>>>>> braille from EPUB or NIMAS files. Some of the techniques planned could
>>>>>> use accurate back translation.
>>>>>> How serious are the back translation problems?
>>>>>>
>>>>>> -----Original Message-----
>>>>>> From: brailleblaster-bounce@xxxxxxxxxxxxx
>>>>>> [mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of Michael
>>>>>> Whapples
>>>>>> Sent: Friday, January 24, 2014 4:49 AM
>>>>>> To: brailleblaster@xxxxxxxxxxxxx
>>>>>> Subject: [brailleblaster] Re: LiblouisUTDML and backtranslate
>>>>>> string
>>>>>>
>>>>>> My understanding was that the project was to build a tool for producing
>>>>>> Braille documents. Maintaining the original text and indexing it with
>>>>>> translated Braille I believe is to remove the need for backtranslation,
>>>>>> whilst allowing synchronisation of text and Braille (eg. for embossers
>>>>>> with ink and Braille, to know where edited Braille is, etc).
>>>>>>
>>>>>> May be things moved on without me knowing and may be backtranslation is
>>>>>> part of the project scope now. If it is may be someone could explain how
>>>>>> it fits in.
>>>>>>
>>>>>> Michael Whapples
>>>>>>> On 24/01/2014 09:43, Larry Skutchan wrote:
>>>>>>> Isn't a reliable back translation one of the premises we build this
>>>>>>> project on?
>>>>>>>
>>>>>>> -----Original Message-----
>>>>>>> From: brailleblaster-bounce@xxxxxxxxxxxxx
>>>>>>> [mailto:brailleblaster-bounce@xxxxxxxxxxxxx] On Behalf Of John J.
>>>>>>> Boyer
>>>>>>> Sent: Thursday, January 23, 2014 11:11 PM
>>>>>>> To: brailleblaster@xxxxxxxxxxxxx
>>>>>>> Subject: [brailleblaster] Re: LiblouisUTDML and backtranslate
>>>>>>> string
>>>>>>>
>>>>>>> The string does not have to be in xml format. That said, try to avoid
>>>>>>> back-translation as much as possible. It is unreliable. Granted, there
>>>>>>> are some bugs, but even under the best of conditions it can give bad
>>>>>>> results. ViewPlus discovered this, and it wºas one reason they proposed
>>>>>>> UTDML.
>>>>>>>
>>>>>>> John
>>>>>>>
>>>>>>>> On Thu, Jan 23, 2014 at 03:26:32PM -0500, Brandon Roller wrote:
>>>>>>>> Does the string that I want to back translate need to be in xml
>>>>>>>> format like translateString? I can't find anything about
>>>>>>>> backtranslateString in the liblouisutdml documentation.
>>>>>>> --
>>>>>>> John J. Boyer; President, Chief Software Developer Abilitiessoft, Inc.
>>>>>>> http://www.abilitiessoft.com
>>>>>>> Madison, Wisconsin USA
>>>>>>> Developing software for people with disabilities
>>
>
>
Other related posts:
