| 英文摘要 |
"This paper examines the applications of copyright protection standards to the creation and utilisation of databases for machine learning purposes and compares the law from Malaysia, Taiwan and the European Union. The current international regime for copyright protection of databases requires that compilations of data be protected “by reason of the selection or arrangement of their contents [which] constitute intellectual creations'. However, Malaysia continues to follow the sweat of the brow approach to copyright protection of tables or compilations although a provision similar to TRIPS separately exists. On the other hand, Taiwan adopts a similar, but not identical approach to the TRIPS. The key differences are that Taiwan’s formulation uses the word “creativity” instead of “creation” sans “intellectual” and uses the conjunction “and” instead of “or” between selection and arrangement. The ECJ’s decision in Football Dataco Ltd v Yahoo! UK Ltd in interpreting the European Database Directive emphasised that copyright protection for databases must be determined solely on the basis of the selection or arrangement but not on the creation of content. Likewise, the Fixture Marketing cases decided by the ECJ held that only the obtaining, verification or presentation of “existing independent material” matters to the sui generis database rights but not that of newly created data. It would appear that most, if not all, machine learning datasets will not satisfy the “intellectual creations” requirement and thus fail to qualify for copyright protection. Enacting a universal sui generis database protection may not be an easy solution. Perhaps, it is time for the international community to get down from the philosophical high horse and to accept that databases should be protected in copyright simply on account of the effort in their compilation, without the necessity to judge whether there is any intellectual input in the selection or arrangement of their content." |
本站僅提供期刊文獻檢索。
