Digitální studovna Ministerstva obrany ČR: využití technologií na pokročilou indexaci obsahu historických dokumentů

• Authors: Marek Fišer , Tomáš Kykal

Title variants

EN

The digital reading Room of the Ministry of Defence of the Czech Republic: using technology for advanced indexing of historical documents

• Languages of publication: CS

Abstracts

EN

Developments in information technology and artificial intelligence are providing tools that have considerable potential to facilitate and enrich research in the fields of history and related sciences. A prerequisite for their effective use, however, is the most perfect conversion of analogue historical sources into machine-readable form, so that the search, classification and extraction of the information contained in them is as efficient as in born-digital sources. In their study, Kykal and Fišer first provide an overview of the development of digital libraries and the making available of the results of digitization in the Czech Republic, taking into account the different strategies and technological backgrounds of libraries and archives. They reflect on the limitations of full-text search and point out a surprising systemic deficit in current digital libraries, namely the absence of the diagnostics of the quality of machine transcription performed by Optical Character Recognition (OCR) programs. They then pay special attention to presenting the parameters and possibilities of the Digital Reading Room of the Ministry of Defence of the Czech Republic (Digitální studovna Ministerstva obrany ČR, DSMO), which is based on the Kramerius Digital Library system. Thanks to its role as an aggregator of the digitization production of the memory institutions of the Ministry of Defence, the Reading Room makes available both library documents and digitized items from archive collections and museum collections. Using the example of a printed periodical of the Austro-Hungarian Army from the First World War, the process of the additional enhancement of OCR results using the PERO tool (Czech abbreviation for pokročilá extrakce a rozpoznávání obsahu - Advanced Extraction and Recognition of Content) is presented, including enrichment with a metadata scheme which captures the layout of graphic and text objects (Analysed Layout and Text Objects, ALTO) and allows the precise localization of the searched text on the digitized image. Using this program, the textual content of not only printed or typewritten texts, but also handwrit­ten texts, can be retrieved much more efficiently and with noticeably higher quality. Moreover, the data in the ALTO scheme could be used to automatically monitor the quality of OCR results. This procedure would significantly increase the usability of semantic search, machine translation, summarization and many other artificial intelligence tools that are yet to be fully deployed in the Czech Digital Library environment.

Keywords

CS

Česká republika; Digitální studovna Ministerstva obrany ČR; archivy; knihovny; digitální knihovna Kramerius; informační technologie; optické rozpoznávání znaků (OCR); sémantické vyhledávání; umělá inteligence; první světová válka; rakousko-uherská armáda

EN

Czech Republic; Digital Reading Room of the Ministry of Defence of the Czech Republic; archives; libraries; Kramerius Digital Library; information technologies; Optical Character Recognition (OCR); semantic search; artificial intelligence (AI); First World War; Austro-Hungarian Army

Discipline

• HISTORY: HISTORY

• Publisher: Institute of Contemporary History, Czech Academy of Sciences
• Journal: Soudobé dějiny (Contemporary History)
• Year: 2024
• Volume: 31
• Issue: 2
• Pages: 447-467
• Document type: ARTICLE

Contributors

author

Marek Fišer

• Soudobé dějiny, redakce, Ústav pro soudobé dějiny AV ČR, v.v.i., Vlašská 9, 118 40 Praha 1, Czech Republic

author

Tomáš Kykal

• Soudobé dějiny, redakce, Ústav pro soudobé dějiny AV ČR, v.v.i., Vlašská 9, 118 40 Praha 1, Czech Republic