Using automatic metadata generation to reduce the knowledge and time requirements for making SCORM learning objects

Abstract

The Internet is packed with resources which can be used for educational purposes, referred to as Learning Objects (LOs). Locating the LO which is best suited for your educational purposes can be extremely challenging. This since the context surrounding the LO in regards to intended user group, educational level etc. are not included in the resource. The SCORM standard has changed this by including contextual metadata as part of the resource. However, SCORM LOs are scarcely created, much as a result of high knowledge and time requirements needed for creating the necessary metadata. This research has been using Automatic Metadata Generation tools to assist in the metadata creation process, enabling LOs to share common contextual metadata while receiving additional high quality LO specific metadata without the need for manual metadata creation efforts.

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

“…the author name(s) should be on the line directly under the title line. This way, AMG efforts based on the Document Code were executed first, and then efforts based on visible characteristics, in line with previous experiences [2,3,8]. A drawback of using AMG algorithms based on visible characteristics is that entities are generated as long as content is located.…”

“…The content of the tagged sections should be retrievable regardless of the document's visible appearance. By lossless converting the Word-documents into the OpenXML Office Word 2007 format, the complete content of the original file's Document Code can be analyzed [2,3,8]. This research created an AMG algorithm to test if the content of the tagged sections actually were retrievable, and usable as a substitute for manual metadata registration efforts while ensuring high metadata quality.…”

“…This can be utilized as AMG strive to avoid excessive manual efforts when similar metadata can be generated automatically based on existing data sources [2,3,4,5,6,7,8]. Previous work have documented that analysis of the document file format, the document code, as seen in Fig.…”

“…Previous work have documented that analysis of the document file format, the document code, as seen in Fig. 1, can reveal extensive amounts of information regarding the document without having to use an interpreter application to re-generate the documents' visual appearance [2,3,8]. These analyses also show that publishers were not willing to generate more metadata then what was system required, often resulting in low quality metadata in the fields filled out.…”

“…This approach uses analysis of the code of the document directly without the need for additional content presentation applications to interpret the document content [2,3,8]. This enables full and direct access to the entire document's content.…”