Get the latest tech news

Compressing Icelandic name declension patterns into a 3.27 kB trie


How I used a compressed trie data structure to encode Icelandic declension data.

The rules are encoded in a trie-like data structure that uses clever compression techniques to get the library’s bundle size under 4.5 kB gzipped. regularity is the degree to which similar key suffixes map to the same values, and comprehensiveness is how well the input data captures rules and exceptions to them. When first developing beygla, I cared a lot about the bundle size being as small as possible so that Icelandic web apps could use the library without being concerned about JavaScript bloat.

Get the Android app

Or read this on Hacker News

Read more on:

Photo of kB trie

kB trie