Computing Reviews

CONTENTUS--technologies for next generation multimedia libraries
Nandzik J., Litz B., Flores-Herr N., Löhden A., Konya I., Baum D., Bergholz A., Schönfu&bgr; D., Fey C., Osterhoff J., Waitelonis J., Sack H., Köhler R., Ndjiki-Nya P. Multimedia Tools and Applications63(2):287-329,2013.Type:Article
Date Reviewed: 08/06/13

This comprehensive discussion of tools and procedures for multimedia digital preservation includes methods for optical character recognition (OCR), music recognition, and semantic analysis methods, among many other ambitious topics. The fully automated workflow the authors describe also provides for metadata creation, quality control, and the presentation of search results. Perhaps most ambitious is the plan to do semantic search, using ontologies and concept recognition. With such an enormous scope, this paper can only sketch the detailed techniques used, but it has ample references to the participating projects. The paper is well written and includes comparisons to other large digital library platforms.

These tools are intended for libraries with collections of analog media, such as paper, vinyl, and tape, that desire to convert these to digital versions with useful metadata descriptions. In addition to advanced methods for the cleanup and organization of converted material, the process can involve semantic analysis on multiple levels. Semantic analysis may involve structuring, such as when breaking a video into scenes, or may extend to include detecting known entities or fitting aspects of content into a defined hierarchical ontology. The tools that do this kind of work go beyond recognition to include logical reasoning based on a knowledge structure that is given to the software. An example of such a knowledge structure is the hierarchical nature of geographical information shown in the GeoNames work.

This research is being extended to a development effort involving several major research institutes and libraries in Germany. The software runs on a cluster with a test collection of approximately 20 terabytes of text, including books, newspapers, paper ephemera, photographs, sound recordings, and moving image material. There is no formal evaluation yet, but if successful, even individual parts of this project should greatly ease the task of converting historical collections to digital formats. The paper suggests that this software will be provided as a commercial product when finished.

Reviewer:  Michael Lesk Review #: CR141432 (1310-0931)

Reproduction in whole or in part without permission is prohibited.   Copyright 2024 ComputingReviews.com™
Terms of Use
| Privacy Policy