Does the Kindle not recognize some diacritics for exact match in dictionary lookups?

[geotadams]

Hi,

I've been slowly going through the process of making my Slovak-German dictionary fully functional with the ability to look up words with inflections, and I've recently run into a problem. Slovak is a language with a great deal of diacritical markings, and the Kindle seems like it may treat some differently than others. The dictionary seems to differentiate between normal accent marks and umlauts just fine, such as áéíóú. But when it comes to some of the more unusual diacritics in Slovak, it doesn't seem to bother. Examples:

Koza - goat, Koža - skin, leather. Although on both of these the dictionary is told to look for an exact match, it always brings up the entry for koza when koža is selected.

Cigána - accusative of gypsie, cigáňa - accusative of liar (admittedly these words are etymologically related thanks to racism, but they should still be treated separately by the dictionary) - selecting cigáňa always brings up gypsie.

So it seems that the "exact match" parameter may only work with certain diacritics, unless I'm missing something here. Does anyone have a suggestion as to how I could fix this problem, or a possible workaround? Below is my code for the examples:

      <idx:entry name="word" scriptable="yes">
        <h2>
          <idx:orth value="cigán">Cigán
		  <idx:infl>
		  <idx:iform value="cigán" exact="yes" /> 
		  <idx:iform value="cigána" exact="yes" /> 
		  <idx:iform value="cigánovi" exact="yes" /> 
		  <idx:iform value="cigán" exact="yes" /> 
		  <idx:iform value="cigánovi" exact="yes" /> 
		  <idx:iform value="cigánom" exact="yes" /> 
		  <idx:iform value="cigáni" exact="yes" /> 
		  <idx:iform value="cigánov" exact="yes" /> 
		  <idx:iform value="cigánom" exact="yes" /> 
		  <idx:iform value="cigánov" exact="yes" /> 
		  <idx:iform value="cigáni" exact="yes" /> 
		  <idx:iform value="cigánoch" exact="yes" /> 
		  <idx:iform value="cigánmi" exact="yes" /> 
		  </idx:infl>   
		  </idx:orth>
        </h2>
Zigeuner	<i>noun</i>	<b>ethn.</b>
      </idx:entry>


      <idx:entry name="word" scriptable="yes">
        <h2>
          <idx:orth value="cigáň [klamár]">cigáň [klamár]
		  <idx:infl>
		  <idx:iform value="cigáň" exact="yes" /> 
		  <idx:iform value="cigáňa" exact="yes" /> 
		  <idx:iform value="cigáňovi" exact="yes" /> 
		  <idx:iform value="cigáňa" exact="yes" /> 
		  <idx:iform value="cigáň" exact="yes" /> 
		  <idx:iform value="cigáňovi" exact="yes" /> 
		  <idx:iform value="cigáňom" exact="yes" /> 
		  <idx:iform value="cigáni" exact="yes" /> 
		  <idx:iform value="cigáňov" exact="yes" /> 
		  <idx:iform value="cigáňom" exact="yes" /> 
		  <idx:iform value="cigáňov" exact="yes" /> 
		  <idx:iform value="cigáni" exact="yes" /> 
		  <idx:iform value="cigáňoch" exact="yes" /> 
		  <idx:iform value="cigáňmi" exact="yes" /> 
		  <idx:iform value="cigáne" exact="yes" /> 
		  <idx:iform value="cigáňov" exact="yes" /> 
		  <idx:iform value="cigáňom" exact="yes" /> 
		  <idx:iform value="cigáne" exact="yes" /> 
		  <idx:iform value="cigáne" exact="yes" /> 
		  <idx:iform value="cigáňoch" exact="yes" /> 
		  <idx:iform value="cigáňmi" exact="yes" /> 
		  </idx:infl> 		  
		  </idx:orth>
        </h2>
Lügner	<i>noun</i>
      </idx:entry>

      <idx:entry name="word" scriptable="yes">
        <h2>
          <idx:orth value="koza1">koza
		  <idx:infl>
		  <idx:iform value="koza" exact="yes" /> 
		  <idx:iform value="kozy" exact="yes" /> 
		  <idx:iform value="koze" exact="yes" /> 
		  <idx:iform value="kozu" exact="yes" /> 
		  <idx:iform value="kozou" exact="yes" /> 
		  <idx:iform value="kôz" exact="yes" /> 
		  <idx:iform value="kozám" exact="yes" /> 
		  <idx:iform value="kozách" exact="yes" /> 
		  <idx:iform value="kozami" exact="yes" /> 
		  </idx:infl>		  
		  </idx:orth>
        </h2>
<p><b>1.</b> Ziege	<i>noun</i>	<b>zool.</b></p>
<p><b>2.</b> [samica] Geiß [österr.] [südd.] [schweiz.]	<i>noun</i>	<b>zool.</b></p>
<p><b>3.</b> [konštrukcia; telocvičné náradie] Bock [Gestell; Turngerät]	<i>noun</i>	<b>sport</b> <b>tech.</b></p>
<p><b>4.</b> [telocvičné náradie] Sprungbock	<i>noun</i>	<b>sport</b></p>
<p><b>5.</b> [kozy] [prsia] Titten [ugs.]	<i>noun</i></p>
      </idx:entry>


      <idx:entry name="word" scriptable="yes">
        <h2>
          <idx:orth value="koža2">koža
		  <idx:infl>
		  <idx:iform value="koža" exact="yes" /> 
		  <idx:iform value="kože" exact="yes" /> 
		  <idx:iform value="koži" exact="yes" /> 
		  <idx:iform value="kožu" exact="yes" /> 
		  <idx:iform value="kožou" exact="yes" /> 
		  <idx:iform value="koží" exact="yes" /> 
		  <idx:iform value="kožiam" exact="yes" /> 
		  <idx:iform value="kožiach" exact="yes" /> 
		  <idx:iform value="kožami" exact="yes" /> 
		  </idx:infl>		  
		  </idx:orth>
        </h2>
<p><b>1.</b> Haut	<i>noun</i></p>
<p><b>2.</b> Leder	<i>noun</i></p>
      </idx:entry>

[jhowell]

I did some limited testing of this and as far as I can tell exact="yes" does nothing, despite being documented in the Amazon Kindle Publishing guidelines.

More details:

I created a short test dictionary using the example of exact="yes" from the Amazon Kindle Publishing Guidelines section 15.4.2 (Exact-match Parameter) and compiled the dictionary using kindlegen 2.9. I then modified the source to remove exact="yes" and compiled it again.

The only differences in the resulting MOBI dictionary files were in metadata that always changes (creation time, modification time, book version, and container ID). There were no differences in the actual content of the file. This indicates that the "exact" attribute is ignored by kindlegen.

I also tried this with Mobipocket Creator 4.2 and had the same result.

[geotadams]

Originally Posted by jhowell

I did some limited testing of this and as far as I can tell exact="yes" does nothing, despite being documented in the Amazon Kindle Publishing guidelines.

More details:

I created a short test dictionary using the example of exact="yes" from the Amazon Kindle Publishing Guidelines section 15.4.2 (Exact-match Parameter) and compiled the dictionary using kindlegen 2.9. I then modified the source to remove exact="yes" and compiled it again.

The only differences in the resulting MOBI dictionary files were in metadata that always changes (creation time, modification time, book version, and container ID). There were no differences in the actual content of the file. This indicates that the "exact" attribute is ignored by kindlegen.

I also tried this with Mobipocket Creator 4.2 and had the same result.

God, that sucks. That makes it much more difficult to make a good dictionary for Slovak as there's a lot of words that are very very different depending on their diacritical markings. One wonders why they would build that in without adding functionality.

[jhowell]

I came across some old documentation that mentioned using the exact="yes" attribute in conjunction with <idx:key>. I tried it in a test file and found that adding it does in fact carry over as a change in the dictionary created using kindlegen.

Unfortunately, it does not appear that this will be useful in practice. Exact seems to do nothing when used with <idx:iform> and the current Kindle publishing guidelines state that <idx:key> is deprecated and should not be used.

[geotadams]

Originally Posted by jhowell

I came across some old documentation that mentioned using the exact="yes" attribute in conjunction with <idx:key>. I tried it in a test file and found that adding it does in fact carry over as a change in the dictionary created using kindlegen.

Unfortunately, it does not appear that this will be useful in practice. Exact seems to do nothing when used with <idx:iform> and the current Kindle publishing guidelines state that <idx:key> is deprecated and should not be used.

Interesting. Does that mean that it definitely wouldn’t work on kindle in practice, or just that they prefer you don’t use it? Because this dictionary is primarily for personal use anyway, not a lot of people need or want a Slovak-German dictionary.

[jhowell]

Originally Posted by geotadams

Interesting. Does that mean that it definitely wouldn’t work on kindle in practice, or just that they prefer you don’t use it? Because this dictionary is primarily for personal use anyway, not a lot of people need or want a Slovak-German dictionary.

I don't know whether or not it would work. I have no idea how you would even use idx:key in a dictionary. You could try it and let us know how it goes.

[geotadams]

Originally Posted by jhowell

I don't know whether or not it would work. I have no idea how you would even use idx:key in a dictionary. You could try it and let us know how it goes.

Ah, I had just assumed you could replace idx:iform with idx:key. I'm obviously still quite an amateur at this.