In databases such as AGRIS, there is large quantity of metadata records, dating back several decades. Most of these records do not have full-text documents available in the originating repositories. However, it has been shown by a paper published in the Dublin Core Conference in 20032 that there is an effort to put resources on the Internet retrospectively. This paper proposes that for scientific and technical information resources, a well-structured and high-quality metadata record contains enough information to find that resource on the Internet, and as a consequence, no additional human labour is needed to create or maintain any links to full-text.