Could Automatic Metadata Generation be a digital solution for speedier and easier document publishing?

Abstract

Enabling efficient retrieval and re-usage of digital documents is a major challenge as many documents on the Internet and on Intranets are poorly described with metadata. Manual generation of quality metadata requires skilled human resources, is costly and time-consuming. As a result, metadata related to the documents are too often insufficient or even incorrect. Automatic Metadata Generation (AMG) algorithms could perform similar metadata generation efforts in seconds without the need for human efforts. Submission of conference proceedings commonly includes specifying an extensive range of metadata. Conference proceedings are based on a specific document template with strict usage regulations making them a prime candidate for AMG efforts. This paper evaluates usage of AMG to generate metadata from papers based the MS Word-based IEEE & ACM conference proceedings templates. This enables this research to evaluate if the templates enable efficient AMG efforts, and if the desired paper content is actually retrieved. As authors might not see value in complying with the templates, actual document content can differ from the template specifications.