METHOD AND DEVICE FOR ANALYZING JAPANESE MORPHEME

Japanese Patent JP3348872

B2

PURPOSE: To make it possible to analyze a morpheme without using a dictionary of a large size by segmenting a morpheme character string from a 'KANAJI' (Chinese character)/'KANA' (Japanese syllabary)-mixed Japanese sentence based upon a morpheme dictionary, allocating a morpheme speech part candidate to the morpheme character string and inspecting parts of speech of adjacent morphemes based upon an adjacent morpheme speech part connection table.
CONSTITUTION: Objective character strings S are extracted, the morpheme of the longest coinident character string segmented based upon an extended dictionary and the speech part candidates of the morpheme are set up (steps 1, 2). When the segmentation is failed, whether the leading character string of the sentences S is 'HIRAGANA's (cursive 'KANA' characters) or non 'HIRAGANA's is judged. When the character stirng is 'HIRAGANA's, the morpheme of the longest coincident 'HIRAGANA' character string is segmented based upon a 'HIRAGANA' adjacent dictionary and a 'HIRAGANA' independent word dictionary and the speech part candidates of the morpheme is set up (step 5). In the case of the non-'HIRAGANA' character string, the segmentation of the same character sort character string is executed and the morpheme speech part candidate of each character sort is set up (steps 6, 7). The speech part candidate is deleted based upon the inspection of connection with the morpheme speech part candidate of the preceding morpheme (step 8).

Masayuki Kameda

JP12685192A

November 20, 2002

April 20, 1992

Click for automatic bibliography generation  

株式会社リコー

G06F17/22; G06F17/27; (IPC1-7): G06F17/27; G06F17/22

JP60225274A

Akichika Takano

Previous Patent: SURFACE MEMBER OF ABSORBENT ARTICLE

Next Patent: OUT-OF-FOCUS DETECTION DEVICE AND OPTICAL HEAD USING IT