[liblouis-liblouisxml] backtranslate: a complex hungarian grammatic related issue with I need resolving longer time period
- From: Hammer Attila <hammera@xxxxxxxxx>
- To: liblouis-liblouisxml@xxxxxxxxxxxxx
- Date: Sat, 15 Dec 2012 08:33:30 +0100
Hy,
Now, basical the hungarian back translation works perfect, for example
the hungarian test harness file all backtranslation tests pass
succesfuly. Unfortunately have a complex hungarian grammatic issue the
backtranslation related, with I need longer time period resolve future
if this is possible. This is not fitting the bug fix release time
period, because I think this is not a bug with Liblouis and not a table
related bug. Unfortunately hungarian language some time a very complex
language.
The problem is following:
The hungarian grade1 braille some duplicated single letters and
duplicated letters need marking equals dot combination. A typical
example is the cscs and ccs letters. Both two letters need marking
146-146 dot combinations. Forward translation this is not producing
problem, but backward translation because not possible determining
absolute sure when need backtranslating the 146-146 dot combination with
cscs or ccs letters, backtranslation some words not always producing
right result.
Look two typical examples:
1. Kulccsal word: end users need seeing braille display with
13-136-123-146-146-1-123 dot combinations, this is right happening. This
word are backtranslated right and resulted word are kulccsal when back
translation happened.
2. Furdancscsiga word: end users need seeing braille display with
124-136-1235-145-1-1345-146-146-24-1245-1 dot combinations, this is
right happening. Backtranslation this word not resulting good word,
resulted word are furdanccsiga, not furdancscsiga. Absolute
understandable why happening this, because not possible easyest
determining the two 146-146 dot combination when need backtranslating
with cscs or ccs letter combinations.
Possible solutions:
1. Need defining experienced backtranslation exceptions a small table
and need adding for example the furdancscs word. This is not a good
solution, because we not possible known all this type words and need
always updating the exception list if detecting a word with producing
backtranslation failure.
2. Need using anyway different dot combinations the cscs letter
combination, but end users final need resulting braille output with
146-146 dot combination the braille output he's braille display. I don't
no this is possible doing or not. For example Liblouis supports a
"hidden dot" feature? For example, if possible defining the cscs letter
combination with 146-5-146 dot combination, final braille output need
cutting the 5 dot combination the second cell. During back translation,
this type dot combinations need resulting cscs dot combinations.
Following solution is not a good solution:
If I temporary added the partword cscs 146-146 dot combination,
furdancscsiga word backtranslated correctly, but for example furdanccsal
word is not. Resulted word this situation is furdancscsal, not
furdanccsal. If this rule I removed, resulted backtranslated word right
furdanccsal word.
Anybody have an useful ydea how can I fix longer time period this issue?
Attila
For a description of the software, to download it and links to
project pages go to http://www.abilitiessoft.com
Other related posts:
- » [liblouis-liblouisxml] backtranslate: a complex hungarian grammatic related issue with I need resolving longer time period - Hammer Attila
