Go SymSpell

Overview

Go SymSpell is a fast and efficient spell-checking and correction library for Go. It implements the SymSpell algorithm with the “symmetric delete” approach, enabling both speed and accuracy. Unlike traditional spell checkers that generate variations of the input word, SymSpell precomputes all possible deletions of dictionary words up to a given edit distance. This allows very quick lookups while keeping correction quality high.

Installation

go mod init your-project
go get github.com/Trendyol/go-symspell

Quick Start

package main

import (
    "fmt"
    "log"
    "github.com/Trendyol/go-symspell/symspell"
    "github.com/Trendyol/go-symspell/verbosity"
)

func main() {
    // Create a new SymSpell instance with default settings
    ss, err := symspell.NewSymSpell()
    if err != nil {
        log.Fatal("Failed to create spell checker:", err)
    }

    // Load dictionary (word frequency_count format)
    _, err = ss.LoadDictionary("dictionary.txt", 0, 1, " ")
    if err != nil {
        log.Fatal("Failed to load dictionary:", err)
    }

    // Get spelling suggestions
    suggestions, err := ss.Lookup("speling", verbosity.Top, 2)
    if err != nil {
        log.Fatal("Lookup failed:", err)
    }

    // Print results
    for _, suggestion := range suggestions {
        fmt.Printf("Suggestion: %s (Distance: %d, Frequency: %d)\n", 
            suggestion.Term, suggestion.Distance, suggestion.Count)
    }
}

Configuration Options

Create a SymSpell instance with custom configuration:

ss, err := symspell.NewSymSpell(
    symspell.WithMaxDictionaryEditDistance(2),  // Maximum edit distance for dictionary
    symspell.WithPrefixLength(7),               // Prefix length for optimization
    symspell.WithIncludeUnknown(true),          // Include unknown words in results
    symspell.WithTransferCasing(true),          // Transfer original casing
    symspell.WithIgnoreNonWords(true),          // Skip non-word tokens
    symspell.WithSplitBySpace(true),            // Enable compound word splitting
)

Available Options

Option	Default	Description
MaxDictionaryEditDistance	2	Maximum edit distance for precomputed deletes
PrefixLength	7	Length of word prefixes for optimization
InitialCapacity	16	Initial dictionary capacity
CountThreshold	1	Minimum frequency threshold for words
DistanceAlgorithm	DamerauOSAFast	Edit distance algorithm
IncludeUnknown	false	Include input word even if not in dictionary
TransferCasing	false	Apply original casing to suggestions
IgnoreNonWords	false	Skip tokens that aren't words
IgnoreTermWithDigits	false	Skip words containing digits
SplitBySpace	false	Split compound words automatically

Dictionary Format

Dictionary files should contain words with their frequencies:

the 1061396
of 593677
to 416629
and 411764
a 409757

Verbosity Levels

Control the detail level of suggestions:

import "github.com/Trendyol/go-symspell/verbosity"

// Top suggestion only
suggestions, _ := ss.Lookup("word", verbosity.Top, 2)

// Closest matches within edit distance
suggestions, _ := ss.Lookup("word", verbosity.Closest, 2)

// All suggestions within edit distance
suggestions, _ := ss.Lookup("word", verbosity.All, 2)