Possibility to NOT use accent marks in SMS?

If you wish something new in Smart Keyboard, here is the place to ask!
Posts: 1
Joined: Mon Jan 02, 2012 2:46 pm
Phone: Motorola Defy+ with Gingerbread 2.3.5

Re: Possibility to NOT use accent marks in SMS?

Post by cascaval »

I've recently installed Smart Keyboard and out of those that I tried it's hands down the best one. However the Czech dictionary using the accented characters simply doesn't suit me as I use non-accented words both in SMSs and emails. Therefore I decided to temporarily solve the problem by populating custom dictionary with all (sic) the words in their non-accented form. It gives solid results when writing using the T9 or compact mode. However, it's far from perfect. Here comes few notes:

1. Word list used - I used a word list that comes with PSPad - the older/smaller version is IMHO based on the dictionary that comes with Mozilla Thunderbird. It contains cca 350000 words. Many of them most of the Czech people have never used and never will. Yet it doesn't contain all the possible forms (declension, conjugation) of some of the common words. The newer/bigger version probably contains all of the words and their forms. There's cca 800000 words and this seems to be a bit too many for Smart Keyboard to handle as there are some lags when typing and the app takes a lot of memory. What we would need is a word list of 10000 commonly used words + all their possible forms. IMHO such a word list is not freely available.

2. Import of the words - Smart Keyboard is not really capable to import such a word list so I simply copied userdic.db and populated the database on my PC. Being no expert on SQLite, I used SQLite Administrator to import the words from a CSV file. It took a couple of hours to import. :roll:

3. Result - as alredy mentioned, the result is quite good for the T9 or compact mode. However, it doesn't give any good results in the full keyboard as the app doesn't know the "weight" of the words and thus it doesn't come up with any good predictions. It probably will once you write a LOT of text but hey then you don't need a wordlist anyway as the app would learn the words itself. The only solution that comes to my mind is to take all the emails that I've written in the past years, parse the words and populate the frequencies table in the userdic.db. I'm not sure I want to spend time doing it.
Post Reply