非漢字搜尋

[pcchen]

*羅馬字搜尋（台語羅馬字，客語羅馬字等） *注音符號搜尋 提供不知道怎麼寫漢字的時候的一種搜尋的方法。

[audreyt]

Think of it like spell correction; given any kind of input, you can have different possible "auto-corrections" that matches dictionary entries; on the other dimension you have frequencies of corrections. Each input has its own grid of matched candidates, sorted by cost of modification on one dimension and frequency by another dimension.

The horizontal row would be the pristine (without edit) matches candidates (if any), and each successive row would have more and more distanced edits with each row corresponding to a small delta such as "ba" into "pa" and "bak" etc, with the phonetic cue on the leftmost column and all the matched candidate words list horizontally next to each other.

Concrete example using pseudo-tai-lo when the user types “Pa” :

<— More frequent — Less Frequent —>

[Pah]  | 爸 百 叭 …
[Bak] | 腹 剝 北 …
[Pa] | …
[Pak] | ...