| 摘要 |
<p>A system (10) for use in character (16) recognition (374, 382) in which a directed graph representing a combination of finite state machines implementing string recognition algorithms for a variety of string categories is used to process string data (12) containing a particular ending subsequence. The ending (18, 70, 108) subsequence includes acceptance (36, 80) information indicating whether (80) the string (12) is acceptable. If so, the ending subsequence (18, 70, 108) also includes information indicating a set of categories (36, 82) including words (204, 206), numbers (208, 212), coumpounds words (210), and so forth. The acceptance (18, 36, 108) information can include a bit indicating the character type of the ending character, and an acceptance (18) data unit which indicates an acceptable string ending. The acceptance (18) data unit can be followed by category (18) data units, each indicating a category, which can be used to obtain a bit vector for a string (12), each bit of which indicates whether the string (12) is in one of the categories.</p> |
