cessen/furigana_gen
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Tweak the learning algorithm.

Browse Source 
It was both too conservative and not conservative enough in different
circumstances.
This commit is contained in:
Nathan Vegdahl 2024-09-11 12:05:57 +02:00
parent 44cb2b8bda
commit 80269e1ff4
2 changed files with 19 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
src/learner.rs
View File

@ -1,8 +1,8 @@
use std::collections::HashMap;
const LEARN_RATE: f64 = 1.0;
const MIN_MAX_DISTANCE: usize = 100;
const MAX_MAX_DISTANCE: usize = 10000;
const LEARN_RATE: f64 = 0.5;
const MIN_MAX_DISTANCE: usize = 10;
const MAX_MAX_DISTANCE: usize = 25000;
#[derive(Debug, Copy, Clone)]
pub(crate) struct WordStats {
@ -33,17 +33,17 @@ impl Learner {
/// Returns the word stats, sorted by how "well known" they are according
/// to the `max_distance` metric.
pub(crate) fn word_stats(&self) -> Vec<(String, WordStats)> {
pub(crate) fn word_stats(&self) -> (usize, Vec<(String, WordStats)>) {
let mut stats: Vec<(String, WordStats)> = self
.stats
.iter()
.map(|(w, s)| (w.clone(), s.clone()))
.collect();
stats.sort_unstable_by_key(|(_, s)| s.max_distance);
stats.sort_unstable_by_key(|(_, s)| (s.max_distance, s.times_seen));
stats.reverse();
stats
(self.words_processed, stats)
}
pub fn record(&mut self, word: &str) {
@ -58,10 +58,8 @@ impl Learner {
return;
}
if distance < stats.max_distance {
stats.max_distance +=
distance.min((stats.max_distance as f64 * LEARN_RATE) as usize);
}
stats.max_distance = stats.max_distance.min(MAX_MAX_DISTANCE);
})

12
src/lib.rs
View File

@ -51,17 +51,21 @@ impl FuriganaGenerator {
Self {
tokenizer: Tokenizer::new(dict),
exclude_kanji: exclude_kanji,
learner: Learner::new(if learn_mode { 5 } else { usize::MAX }),
learner: Learner::new(if learn_mode { 3 } else { usize::MAX }),
}
}
pub fn word_stats(&self) -> Vec<(String, usize, usize)> {
let mut stats = self.learner.word_stats();
/// Returns (total_words_processed, Vec<(Word, distance, times_seen)>)
pub fn word_stats(&self) -> (usize, Vec<(String, usize, usize)>) {
let (total_words, mut stats) = self.learner.word_stats();
(
total_words,
stats
.drain(..)
.map(|(w, s)| (w, s.max_distance, s.times_seen))
.collect()
.collect(),
)
}
pub fn add_html_furigana(&mut self, text: &str) -> String {