OCR : Unleash the hidden information

Abstract

Most of us, even though it is not very rational, commonly take pictures of texts. In a conference it is very unlikely not to see participants taking pictures of presentation slides. Similarly, national archives scan documents without doing an OCR (Optical Character Recognition). Resulting image, in spite of its resolution, quality or file format is not searchable by its content. Unless someone types in a large amount of metadata according to Dublin Core for example. While this is an acceptable behavior in an archival world, an average people is willing to fill the maximum of five fields. Therefore a clear need for an easy and most importantly a free way to get pictures, scanned documents etc. to be fully searchable is a mandatory need.

A Digitalia research center has been working on to create an effective workflow that automatically analyzes the document content, generates OCR information as well as gets the most relevant keywords for the content. Furthermore, the workflow produces an archival graded PDF/A file if requested by the user. This workflow has been fully integrated into our Citizen Archive solution to handle everything automatically in the background. With this sophisticated solution usability, findability as well as reusability of the preserved content will be greatly increased. In short this equals better archival user experience and less manual work to be done for both the archivist and the end user.