Re: [Tracker] Additions to Tracker Ontology for Chinese Use Cases

[Aleksander Morgado <aleksander lanedo com>]

> > Here are some examples showing "name in Chinese" -> "name in pinyin" -> "T9 representation":
> >
> > åé  ->  Liu Jing  ->  5485464
> > çå  ->  Luo Xuan  ->  5869826
> > éèè  ->  Ma Fang Fang  -> 6232643264
> >
> > The first use case is that it is apparently very common for phone users in China to look up names in 
> > their contact list using the T9 dial pad. So in data from above, if the user clicks '5' on the dial pad, 
> > they want to perform a search using Tracker to get the results "Liu Jing" and "Luo Xuan". When the user 
> > then clicks the '8' on the dial pad Tracker would be searched again and only return "Luo Xuan".
> >
> > My customer also wants to support searching for Chinese contacts using their Pinyin names. So if the user 
> > types "L" using the virtual or physical keyboard on the phone, again both "Liu Jing" and "Luo Xuan" 
> > should be returned. If the user adds an "i", another Tracker search will be performed which should return 
> > only "Liu Jing".
> >
> > The other important use for the Pinyin name is for sorting the contacts list. My customer wants to sort 
> > Chinese names among English names using their Pinyin representation. So if the English Names "Matt 
> > Compton" and "John Doe" are added to the list of Chinese names above, the list of names should be sorted 
> > alphabetically as follows:
>
> Please consider that when using libicu for unicode support, it will
> allow setting "pinyin" as a collation specifier in the locale being used
> by Tracker. This should, in theory (didn't test it), allow ordering the
> results of the queries based on the pinyin representation of the input
> string. That should also allow looking for Chinese contacts using the
> Pinyin representation of the original chinese name. Anyway, I'm not sure
> how the T9 representation would work here. You could give this a try,
> because it may be enough for your needs (I might be wrong, of course).

Oh, forgot to paste the link of ICU locale information:
http://userguide.icu-project.org/locale

And the UTS#35, which shows which are the possible collation specifiers
suggested by Unicode:
http://www.unicode.org/reports/tr35


-- 
Aleksander