文字探勘在判決書上之應用-以著作權法民事賠償為中心

Abstract

著作權法第六章中規定了著作權人權利受到侵害時之救濟，其中第85條及第88條分別規定了著作人格權及著作權財產權受到侵害時，侵害人所應負之民事責任。但因著作權具有無體財產之特性，所受損害難以證明，加上著作權案件之情況及種類皆繁雜，使得影響損害賠償金額之因子具有相當之不確定性，不利於裁判之進行及訴訟之和解。 本研究以智慧財產法院成立以來四年間之案件為觀察之標的，進行法實證研究分析，統計出現於著作權民事及刑事附帶民事案件判決中，影響損害賠償金額之各項因子，並以文字探勘之概念，結合正規表示式之演算法，設計實驗檢索系統協助實證研究之進行。實驗程式抽樣測試之平均求全率為97.96% ，求準率為95.4%，能有效提高研究工作之效率。

Copyright Act formulates remedies for infringement of rights in the sixth chapter. Article no. 85 and 88 prescribe that a person who infringes on moral rights of an author or economic rights of the copyright holder shall be liable for damages. Because copyright has the features of non-physical properties and copyright cases are quite various, it is difficult for the injured party to prove actual damages. Such complicated situation makes the uncertain factors exist and influences the amount of compensation. It will become disadvantages of the trial process and reconciliation established. This dissertation processes legal empirical studies by using statistics the factors that influence the amount of compensation from the copyright civil trial cases and ancillary civil action of Intellectual Property Court for four years since the establishment. We implement the concept of text mining and combine with regular expressions to design the algorithms of our experimental retrieval program to provide help for the legal empirical studies. The average recall of Experimental program is 97.96% and the precision is 95.4%. The Experimental system is verified to improve the efficiency of research works.