Method and system for morphologizing text

摘要

A method and system for morphologizing written or printed texts, including Japanese texts are obtained in accordance to codes. The longest morphemes are divided one at a time from the characters in a sentence. This is achieved by forming the longest morpheme from the remaining characters in the sentence which is listed in a dictionary of valid morphemes and determining if it is conjunctive with the previously divided morpheme. To determine if a formed morpheme is conjunctive, associated pairs of front and back connection codes are retrieved. If a front connection code of one retrieved pair and a back connection code of a pair of connection codes of the previously divided morpheme are co-listed in a table of permissible relationships, the formed morpheme is conjunctive. If no character may be divided from the remaining characters in the sentence, a previously divided morpheme is redivided. If a morpheme can be divided and is conjunctive with the previous morpheme, a connection action, describing the relationship between the formed morpheme and the previously divided morpheme, is recorded. In response to certain connection actions, the next morpheme is divided by forming it from a single character of the remaining characters and testing it. After all of the morphemes are divided, a word graph is constructed from the morphemes in accordance with the connection actions relating adjacent morphemes.