The service provides web access to the MeCab language analysis tool, which breaks down Japanese texts into parts of speech and other morphological elements. It seeks to analyze a text passage based on a general-purpose dictionary and reference collection to reveal structure and word use embedded in the sample. It builds on a Markov model of word sequence, but attempts to improve analytical performance by introducing randomized elements. An open-source installed version is available.
API methods support submission of a Japanese-language text passage (assumed to be a complete sentence) with filters to select only nouns or only unique words. Methods return analysis and breakdown of the passage according to its word choice, parts of speech, implied inflection, and related characteristics.