Kindle Hungarian Dictionary

[bubak]

Hi all,

I'm trying (again) to make a Hungarian/English (German) dictionary for my Kindle voyage. I've taken all public Hungarian dictionaries that I was able to get in text form, about 12 of them, I've cleaned, deduplicated and merged them into one text file and produced the corresponding Kindle dictionary. So far so good. As long as the word in the book appears in the dictionary form, it is correctly translated. Unfortunately, Hungarian is not a language well known for its lack of inflections, the opposite is the case. A Hungarian substantive can have up to 5 distinct slots for suffixes and each of them can be occupied by one of a dozen different suffixes, so the number of suffix combinations makes any explicit enumeration virtually impossible. The same holds for verbs. In my previous attempts I've chosen a couple of suffixes (e.g. accusative and plural for substantives, past tense 3rd persons for verbs) and generated them to each dictionary word but this turned out to be as unusable as without them, the book words tend to always appear in a form that is missing in the dictionary.

One could maybe live with it if it were possible to open the dictionary easily and lookup the base form of the word. Unfortunately, Kindle masters chose to do it the hard way: If the selected word is in the dictionary, then there is a menu point "Open the dictionary". When, however, the word is not found, this menu point is disabled, meaning "When I haven't found the translation, then you would not either". So when I really want to lookup the word, I have to mark some other word which is in the correct dictionary form, display its translation, open the dictionary and only then search for the word I need. Beautiful.

Now to the point: Kindle dictionary has the element <idx:infl> which can list inflections of a given word. This is quite useful for languages like English which do not use many inflections, but obviously not suitable for Hungarian. In the <idx:entry> element documentation they write "The spell attribute enables wildcard search" which seems to suggest that the dictionary is able to do a wildcard search, so it might be possible to use e.g. the orth form "barát*" which would then find the translation of the base word "barát" for all various forms like "barátot", "barátok", "barátom", "barátoméit" etc. Undocumented, there also seems to be the attribute "wild" which suggests the same, but I've tried to use a lot of possible combinations of the attributes and the base orth, to no avail. Did anyone succeed to make the wildcard search work for Kindles?

[Nyarlathopian]

It seems that the wildcard search function is totally up to the Kindle device. So it may or may not work on a Kindle Voyage. Anyway, it shouldn't matter when it comes to formatting the dictionary, other than setting the spell attribute to "yes".

When it comes to Hungarian, you have to decide which words to list as inflections that belong to an encompassing headword. Or list the words in a separate entry. The goal should be to list every single word in the dictionary. Despite the many possible combinations. The number of combinations is still limited. Maybe you also want to consider word frequency if you can find data about it.

[bubak]

Great idea! I've added about 1M inflections from a word frequency dictionary and it works like a charm. Interestingly, the size of the dictionary went from 17M to only 28M and there is no visible speed penalty. So, apology to the Kindle masters, the current infl mechanism seems smarter than I thought.