How to create your own mobipocket dictionary for any language: - Page 11

Zylbath · 12-20-2015, 02:37 PM

Thank you for your answer.

Really? I tried to export it from excel with tab delimited. Strange.

I added the .zip with the four excel files. The most important one would be the Greenlandic_English one. Could you, or someone, convert that into a tab .txt format and I can try again to convert it into a useable dictionary? Hopefully it'll work then. (I know the file has some entries that won't be readable because they are inconsistenly formatted, but that's fine. At least some are useable.)

The problem though is that Mobireader Creator doesn't have the language code for Greenlandic (Kalaallisut). I honestly don't even wonder. I always took English (Canada) (It's the nearest to Greenland ^^).

I really appreciate your help. Thanks a lot!

EbokJunkie · 12-20-2015, 02:56 PM

Converted your txt file to utf8 and applied tab2opf with option -utf.
See attached.

Zylbath · 12-20-2015, 03:21 PM

Hey, thank you, thank you, thank you!

It works, I am fascinated. Of course, morphological complexe words often can't be find (Greenlandic is a polysynthetic language where not rarely 12 morphemes with objects, attributes etc. are combined in one word. I guess a dictionary couldn't master that.) But the shorter words can now be searched. I really thank you!

I wish you all Merry Christmas.
Greetings,
Kevin

EbokJunkie · 12-20-2015, 04:39 PM

BTW, Greenlandic language code is kal or kl.
See here.
You can edit opf file and recreate mobi.

Doitsu · 12-20-2015, 06:20 PM

Quote:

Originally Posted by Zylbath

Hey, thank you, thank you, thank you!

It works, I am fascinated. Of course, morphological complexe words often can't be find (Greenlandic is a polysynthetic language where not rarely 12 morphemes with objects, attributes etc. are combined in one word.

I'm not familliar with Greenlandic morphology, but if it has somewhat predictable patterns, you might be able to add inflections.
E.g. if an hypothetical Greenlandic word ABCD will often occur with a prefix aaa and/or a suffix bbb, you could add them as inflections (aaaABCD, ABCDbbb, aaaABCDbbb etc.).
For more information on inflections, see the Kindle Publishing Guidelines and the Mobipocket website.

Quote:

Originally Posted by EbokJunkie

BTW, Greenlandic language code is kal or kl.
See here.
You can edit opf file and recreate mobi.

Unfortunately, the language codes in kindlegen/mobigen are hardcoded, and if you use a language code that is not in the following list, you'll get an error message.

Spoiler:

Zylbath · 12-20-2015, 06:38 PM

Quote:

BTW, Greenlandic language code is kal or kl.
See here.
You can edit opf file and recreate mobi.

Yeah, I tried it with kl only and it resulted in an error. But with kal it worked now, thanks.

Quote:

I'm not familliar with Greenlandic morphology, but if it has somewhat predictable patterns, you might be able to add inflections.
E.g. if an hypothetical Greenlandic word ABCD will often occur with a prefix aaa and/or a suffix bbb, you could add them as inflections (aaaABCD, ABCDbbb, aaaABCDbbb etc.).
For more information on inflections, see the Kindle Publishing Guidelines and the Mobipocket website.

That would be a lifework. Greenlandic only has suffixes, but therefore 1,500 of them, plus the 4,000 stems they can be attached to. And the other things is that many suffixes swallow the consonants before them or they assimilate to just one lengthened consonant, so the word gets unrecognisable changed. Well, not totally, otherwise the language couldn't be understood. But I guess it would be a lifework to do such a dictionary. The dictionary you converted for me already has thousands of inflected forms. But it is just by chance that some words you want to look up really match these, there are just too many combinations with 1,500 suffixes and words that not rarely have 12 morphemes or more. But shorter words are recognised and the dictionary is actually pretty useful. So, thanks again!

Doitsu · 12-20-2015, 07:11 PM

Quote:

Originally Posted by Zylbath

Yeah, I tried it with kl only and it resulted in an error. But with kal it worked now, thanks.

Actually, kal did not work. KindleGen/Mobigen will ignore the third letter and treats kal like ka = Georgian.

Quote:

Originally Posted by Zylbath

That would be a lifework. Greenlandic only has suffixes, but therefore 1,500 of them, plus the 4,000 stems they can be attached to.

Google Open Source Greenlandic POS Tagger Stemmer. Maybe one of the projects has a list containing the headword and all possible inflected forms, because such a list is required for some of the simpler stemmers to reduce inflected forms to headwords. (Usually FST stemmers don't come with human-readable word databases.)

Quote:

Originally Posted by Zylbath

So, thanks again!

I didn't do anything. EbokJunkie converted your file for you.

Zylbath · 12-20-2015, 07:37 PM

It did not work? Damn. But then it has be Georgian, nobody reads in Georgian. =P

There was once a page where the words were analysed automatically and put into a morphological tree. But I can't remember the site, all I know is this: http://giellatekno.uit.no/cgi/d-kal.eng.html It can disambiguate or hyphenate the input. But I don't know whether they have a database that could be used, most of their functionality doesn't work anymore, only some analysis tools are available anymore.

But I can't really find something by the key words you gave. There is something like that for Inuktitut, closely related to Greenlandic, but every morpheme had to be changed because of sound changes.

Teom@n · 01-09-2017, 12:55 PM

tab2opf seperated my txt file to 4 pieces of html. but it doesn't use the UTF-8. I have a problem with special letters. How could I solve this problem?

I tried different methods (calibre, etc) for creating html to use mobigen/kindlegen but this time my kindle doesn't see my dict as dict.

Doitsu · 01-09-2017, 01:28 PM

Quote:

Originally Posted by Teom@n

tab2opf seperated my txt file to 4 pieces of html.

That's the normal behavior. It'll split up large files into several smaller files.

Quote:

Originally Posted by Teom@n

[...] but it doesn't use the UTF-8. I have a problem with special letters. How could I solve this problem?

You'll need to specify the -utf8 command line parameter:

Code:

python tab2opf.py -utf8 dict.txt

Quote:

Originally Posted by Teom@n

I tried different methods (calibre, etc) for creating html to use mobigen/kindlegen but this time my kindle doesn't see my dict as dict.

You can't generate dictionaries with Calibre, you'll need to use Kindlegen. The source files also need to be formatted according to the Kindle Publishing Guidelines. I.e., you'll have to post-edit the files generated by tab2opf.py.

Out of curiosity: what are the input and output languages?

Teom@n · 01-09-2017, 01:34 PM

Quote:

Originally Posted by Doitsu

That's the normal behavior. It'll split up large files into several smaller files.

You'll need to specify the -utf8 command line parameter:

Code:

python tab2opf.py -utf8 dict.txt

You can't generate dictionaries with Calibre, you'll need to use Kindlegen. The source files also need to be formatted according to the Kindle Publishing Guidelines. I.e., you'll have to post-edit the files generated by tab2opf.py.

Out of curiosity: what are the input and output languages?

Firstly, Thanks.

Could you explain "how should I use pyhton code?

Teom@n · 01-09-2017, 01:39 PM

Quote:

Originally Posted by Doitsu

That's the normal behavior. It'll split up large files into several smaller files.

You'll need to specify the -utf8 command line parameter:

Code:

python tab2opf.py -utf8 dict.txt

You can't generate dictionaries with Calibre, you'll need to use Kindlegen. The source files also need to be formatted according to the Kindle Publishing Guidelines. I.e., you'll have to post-edit the files generated by tab2opf.py.

Out of curiosity: what are the input and output languages?

I'm trying to make a French-Turkish dictionary. My tools are:

-tab2opf.exe
-mobigen.exe

-kindlegen
-notepad++

python 3 is intalled on my pc.

edit:
I installed python 2.7 and converted my dict. It works. Now I have to solve hypenation. How should I edit and optimize my database for the best search result? I have a problem with plural words as "inédit, inédite, inédites" and some cells contains a few words as "lequel, laquelle, lesquels, lesquelles ".

My databes is like:

aboyant, ~e*
aboyer*
aboyeur, ~euse*
abracadabrant, ~e*
abraser*
abrasif, ~ive*
abreuvement, abreuvage*
abréviatif, ~ive*
abricoté, ~e*
abricotier*
.
.
.
lequel, laquelle, lesquels, lesquelles

Doitsu · 01-10-2017, 04:46 AM

Quote:

Originally Posted by Teom@n

I installed python 2.7 and converted my dict. It works. Now I have to solve hypenation. How should I edit and optimize my database for the best search result? I have a problem with plural words as "inédit, inédite, inédites" and some cells contains a few words as "lequel, laquelle, lesquels, lesquelles ".

You'll need to explicitly define them as inflections. E.g.

Code:

<html>
<body>

<idx:entry>
	<b><idx:orth>cheval
	<idx:infl>
		<idx:iform value="chevaux"/>
	</idx:infl>
	</idx:orth> </b> 
	beygir 
</idx:entry>

<hr/>

<idx:entry>
	<b><idx:orth>feu
	<idx:infl>
		<idx:iform value="feux"/>
	</idx:infl>
	</idx:orth> </b> 
	ateş, yangın
</idx:entry>

</body>
</html>

For French verbs you might find the dsl2mobi French verb inflection list helpful.

Teom@n · 01-10-2017, 08:40 AM

Quote:

Originally Posted by Doitsu

You'll need to explicitly define them as inflections. E.g.

Code:

<html>
<body>

<idx:entry>
	<b><idx:orth>cheval
	<idx:infl>
		<idx:iform value="chevaux"/>
	</idx:infl>
	</idx:orth> </b> 
	beygir 
</idx:entry>

<hr/>

<idx:entry>
	<b><idx:orth>feu
	<idx:infl>
		<idx:iform value="feux"/>
	</idx:infl>
	</idx:orth> </b> 
	ateş, yangın
</idx:entry>

</body>
</html>

For French verbs you might find the dsl2mobi French verb inflection list helpful.

I have official latest ODS7 and I converted it to Excel. (415.000 words)
I will try to fill it with Turkish explanations than convert stardict-text-html-mobi.
What do you think? Could it be better?

Doitsu · 01-10-2017, 11:23 AM

Quote:

Originally Posted by Teom@n

I have official latest ODS7 and I converted it to Excel. (415.000 words)
I will try to fill it with Turkish explanations than convert stardict-text-html-mobi.
What do you think? Could it be better?

That depends on your technical skills. If your MS Excel spreadsheet contains only two columns you could save it as a tab-delimited text file and process it with tab2opf.py.
If you're familiar with regular expressions, you could also convert the tab-delimited text file directly using a couple of regular expressions.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Dictionary lookup in iBooks 1.1: "Dictionary not available for this language"	kjk	Apple Devices	71	09-18-2010 06:24 AM
best foreign language & dictionary options?	joedevivre	Which one should I buy?	2	12-13-2009 09:40 AM
How do I create headword-enabled Mobipocket dictionary?	owl123	Kindle Formats	1	07-24-2009 11:13 AM
Useful tip: How to change the BD language AFTER you create a book	HarryT	Workshop	4	04-15-2009 12:36 AM
creating a foreign language dictionary	dirtylc	Amazon Kindle	1	03-30-2009 08:40 AM

12-20-2015, 03:21 PM	#153
Zylbath Junior Member Posts: 5 Karma: 10 Join Date: Dec 2015 Device: Kindle	Hey, thank you, thank you, thank you! It works, I am fascinated. Of course, morphological complexe words often can't be find (Greenlandic is a polysynthetic language where not rarely 12 morphemes with objects, attributes etc. are combined in one word. I guess a dictionary couldn't master that.) But the shorter words can now be searched. I really thank you! I wish you all Merry Christmas. Greetings, Kevin

12-20-2015, 04:39 PM	#154
EbokJunkie Addict Posts: 229 Karma: 13495 Join Date: Feb 2009 Location: SoCal Device: Kindle 3, Kindle PW, Pocketbook 301+, Pocketbook Touch, Sony 950, 350	BTW, Greenlandic language code is kal or kl. See here. You can edit opf file and recreate mobi.

12-20-2015, 07:37 PM	#158
Zylbath Junior Member Posts: 5 Karma: 10 Join Date: Dec 2015 Device: Kindle	It did not work? Damn. But then it has be Georgian, nobody reads in Georgian. =P There was once a page where the words were analysed automatically and put into a morphological tree. But I can't remember the site, all I know is this: http://giellatekno.uit.no/cgi/d-kal.eng.html It can disambiguate or hyphenate the input. But I don't know whether they have a database that could be used, most of their functionality doesn't work anymore, only some analysis tools are available anymore. But I can't really find something by the key words you gave. There is something like that for Inuktitut, closely related to Greenlandic, but every morpheme had to be changed because of sound changes.

01-09-2017, 12:55 PM	#159
Teom@n Enthusiast Posts: 47 Karma: 10 Join Date: Dec 2014 Location: Lyon Device: Kindle PW3, Kobo Libra H2O	tab2opf seperated my txt file to 4 pieces of html. but it doesn't use the UTF-8. I have a problem with special letters. How could I solve this problem? I tried different methods (calibre, etc) for creating html to use mobigen/kindlegen but this time my kindle doesn't see my dict as dict.

Advert

Advert