PURPOSE: To reduce the burden of resolution of various meanings and nouns in a machine translator by collecting uppercase letter start tokens into one noun phrase in the English morpheme analysis at the time of continuous appearance of these uppercase letter start tokens in an English text to accurately analyze a proper noun of a person's name, an enterprise's name, or a group name without dividing it.
CONSTITUTION: A character string dividing part 2 divides an inputted alphabet string into character strings as units of dictionary retrieval. A token classification recognizing part 3 determines the classifications of tokens by characters constituting the tokens divided by the character string dividing part. A dictionary retrieval part 5 retrieves a dictionary part 4 with individual divided tokens as keys to acquire retrieval information. When uppercase letter start tokens continuously appear, an uppercase letter start token string recognition part 6 collects them into one. A person's name pattern recognition part 8 collates the token string recognized by the uppercase letter start token string recognition part 6 with contents of a person's name pattern table 7 to recognize the token string coinciding with a person's name pattern.
JP6390488 | Document creation support device, program and document creation support method |
WO/2008/155309 | METHOD OF TEXT TYPE-AHEAD |
JPH096777 | WORD PROCESSOR |
Next Patent: DOCUMENT/SENTENCE KNOWLEDGE STORAGE DEVICE