#nlp #frequencies #word #port #up #language #look

wordfreq

Yet another Rust port of wordfreq for looking up the frequencies of words in many languages

6 releases

0.2.3 Jun 24, 2023
0.2.2 Jun 24, 2023
0.1.1 Jun 4, 2023

#968 in Text processing


Used in 2 crates

MIT/Apache

350KB
13K SLoC

wordfreq

This crate is a yet another Rust port of wordfreq, allowing you to look up the frequencies of words in many languages.

Documentation

https://docs.rs/wordfreq/

Licensing

Source code is licensed under either of

at your option.


lib.rs:

wordfreq

This crate is a yet another Rust port of Python's wordfreq, allowing you to look up the frequencies of words in many languages.

Note that this crate provides only the algorithms (including hardcoded standardization) and does not contain the models. Use wordfreq-model to easily load distributed models. We recommend to see the documentation for quick start.

How to create instances without wordfreq-model

If you do not desire automatic model downloads using wordfreq-model, you can create instances directly from the actual model files placed here. The model files describe words and their frequencies in the text format:

<word1> <freq1>
<word2> <freq2>
<word3> <freq3>
...

You can create instances as follows:

use approx::assert_relative_eq;
use wordfreq::WordFreq;

let word_weight_text = "las 10\nvegas 30\n";
let word_weights = wordfreq::word_weights_from_text(word_weight_text.as_bytes())?;

let wf = WordFreq::new(word_weights);
assert_relative_eq!(wf.word_frequency("las"), 0.25);
assert_relative_eq!(wf.word_frequency("vegas"), 0.75);
assert_relative_eq!(wf.word_frequency("Las"), 0.00);

Standardization

If you want to standardize words before looking up their frequencies, set up an instance of Standardizer.

use approx::assert_relative_eq;
use wordfreq::WordFreq;
use wordfreq::Standardizer;

let word_weight_text = "las 10\nvegas 30\n";
let word_weights = wordfreq::word_weights_from_text(word_weight_text.as_bytes())?;

let standardizer = Standardizer::new("en")?;
let wf = WordFreq::new(word_weights).standardizer(standardizer);
assert_relative_eq!(wf.word_frequency("Las"), 0.25); // Standardized

Precision errors

Even if the algorithms are the same, the results may differ slightly from the original implementation due to floating point precision.

Unprovided features

This crate is a straightforward port of Python's wordfreq, although some features are not provided:

Dependencies

~7MB
~131K SLoC