Automatic Hyperlink Generation For Chinese Documents

• Main Authors: Huang Chen-Chu, 黃承渠
• Other Authors: Joung Yuh-Jzer
• Format: Others
• Language: zh-TW
• Published: 1998
• Online Access: http://ndltd.ncl.edu.tw/handle/01790402579067243244

碩士 === 大同工學院 === 資訊工程研究所 === 86 === With the rapid growth of digital documents, the need for a friendly and efficient browsing environment is getting obvious. For English documents, automatic hypertext conversion has stepped into application area. But for Chinese documents, there are still some issues have to be solved before we can ahead the road. The thesis proposes a method to convert existing Chinese news documents into hypertext by add HTML anchor tag on keywords. When users browse hypertext version news, they can query related document simply click on the keyword. The system do a premier statistical approach word identification by looking up a general term dictionary. Then using memorized and recursive algorithm to discover new terms. At last, calculating term weight by combining some filter rule to emphasis prop nouns.For increasing system performance, we propose an efficient way to approximately calculate term weight. It uses an auxiliary table to store term distribution information and decrease the overhead of repeatedly calculating it each time a new document arrives.The experiment result shows that the proposed method does improve system performance and maintain hyperlink quality at the same time. Thus we conclude that it can be adapted for an incremental document collection.