Case Study: Retrieving Scanned Library Cards

Abstract

In this chapter we present a case study where the speech retrieval approach described in is applied to retrieve scanned library cards. Some concepts of are repeated in this chapter such that it is self-contained. The library catalogue of this case study is the Alphabetischer Zentralkatalog of the Zentralbibliothek Zürich. It contains many old, and sometimes even handwritten cards. As a consequence, the Optical Character Recognition (OCR) produces many recognition errors. Since library cards are rather short documents, we cannot expect that the effects of recognition errors cancel out (Section 5.6). Fortunately, we have an additional source of information, i.e. the alphabetic ordering of the cards. We will show how retrieval is improved by taking advantage of this ordering information.