The study on the identification of null elements in natural language processing

• Main Authors: Tsuei,Wen, 崔文
• Other Authors: Chen,Hsin-Hsi
• Format: Others
• Language: zh-TW
• Published: 1996
• Online Access: http://ndltd.ncl.edu.tw/handle/88725353706754484466

碩士 === 國立臺灣大學 === 資訊工程研究所 === 84 === Null elements are very important in natural language processing. It affects many natural language applications,such as parsers, machine translations and anaphora resolutions. We want to predict the nulls for each position between words in a sentence. If the position has a null, we must say *yes*, and for the position without nulls, we should say *no* instead. A corpus -based approach is used here. We do the clustering on the training corpus, and induce some patterns as the rules of prediction in these classes. We propose a new way to cluster the sentences with various lengths. The patterns are extracted from the clustered sentences by using the Longest Common Sub- sequences. It is also a new idea, and is different from the original pattern matching methods. We have a 76.74% recall and 70.36% precision on predictions of null elements at last.