Our search algorithm is a proprietary algorithm we call Jgram. It splits search inputs into a contiguous sequence of items from a given sample of characters, and can be words, syllables, sounds or numbers. While other vendors may use one algorithm for sound “soundex” and others for comparing whole text “Levenshtein” our search algorithm examines all facets of a match.
After you provide input, we normalize text to ensure that items like capitalization, punctuation and accented characters do not affect the score, while accounting for common challenges like gluing and reordering of words. Jgram then splits text into variable length tokens and takes the ratio of tokens that match to come up with a similarity score. Finally, Jgram highlights the relevant part of the match in red (if you're using our platform).