摘要 |
PURPOSE:To accurately process a compound word, etc., by processing key word extraction not by the unit of a single word but by the pattern of consecu tive single words. CONSTITUTION:At morpheme analysis (step 1), an input text is morpheme- analyzed and divided into single words to judge a part of speech by each single word. At pattern collation (step 2), this method collates the consecutive part of word coincident with a part of speech pattern (an extracted word pattern) describing the consecutive part of the words which should be extracted as a key word. The method extracts the consecutive part of the single words coincident with single word pattern which is previously defined from the single word string obtaind by the morpheme analysis of a Japanese document like this. |