User:BHaslhofer

From Open Annotation Collaboration

Contents 1 Bernhard Haslhofer 2 Use Case: Historic Map Annotations with YUMA 2.1 Annotating Historic Maps with YUMA 2.2 Target Collections and Audience 2.3 Project Background and Organizational Issues 2.4 References

Bernhard Haslhofer

Research Group Multimedia Information Systems

Faculty of Computer Science

University of Vienna

Use Case: Historic Map Annotations with YUMA

Annotating Historic Maps with YUMA

Historic maps are a valuable scholarly resource. They record historical geographical information often retained by no other written source [9], and are thus relevant to the study of a range of environmental, ecological or socio-economic phenomena: from the development of land use [6], [7] to the effects of river channel changes [2] or floods [10], to the reconstruction of past urban environments [5]. At the same time, they capture more than mere geographic facts: they also draw a fascinating picture of the cultural, political and religious context in which they were created. Their degree of accuracy tells much about the state of technology and scientific understanding at the time of their creation [9]. Consequently, historic maps are cultural heritage artifacts in their own right, part of the artistic heritage as much as of the history of science and technology as a whole [1].

With our YUMA Map Annotation Tool we provide, besides common annotation functionalities, two features that are not satisfied by any other existing tools and services:

• Semantic augmentation of textual annotations with relevant information from the Linked Data Web. The system analyzes the annotation text using named entity recognition. If, for example, people or places are mentioned in the text, the system will suggest links to appropriate resources (e.g. from DBpedia) and prompt the user to verify this information [4].
• Georeferencing. The system allows scholars to establish a georeference for the map (i.e. a correspondence between the map's image coordinates and real-world geographical coordinates) by placing so-called control points - points on the map where the geographical location is known [8]. If the map has been georeferenced with control points, the system can translate between annotated areas and geographic locations. Furthermore, the system can suggest geographical entities (e.g. from Geonames) that intersect with the annotated map region. For example, if a scholar annotates the area of Yosemite National Park on a map of California (see Figure 1), the system will suggest the country (United States) and relevant cities in the area such as Oakhurst, Yosemity Lakes, or Mariposa.

The produced annotation data are exposed as Linked Data on the Web following the LEMO model, which is an extension of the W3C Annotea vocabulary (see [3]). However, the complete system is currently being re-designed and open-sourced ([1]) and new annotation models -- in particular the OAC model -- are investigated. Major technical challenges to be solved w.r.t OAC include: how to represent linked resources as part of map annotations; how to express different types of semantic relationships between annotation/annotated media and resources; how to model a map fragment (which represents a fragment in a digital image and a geographical area at the same time); and how to include georeferencing information as part of a map annotation.

Figure 1: YUMA Map Annotation Tool Screenshot.

Target Collections and Audience

The YUMA Map Annotation Tool relies on zoomable Web image formats (such as Zoomify or Map Service Speci�cation TMS) in order to display high-resolution map scans in the Web browser. However, the system is technically open for any type of map material. If no zoomable version of the map image is available, the system can perform on-the-fly conversion from a wide range of source image formats. At the moment we are conducting a larger experiment with the Library of Congress' Map Collection where approx. 6000 map scans in JPEG 2000 format were processed using YUMA's integrated conversion functionality.

We are targeting scholars from the digital library and map history domain, but also non-expert users from the Web. In a side-experiment, where we aim at creating a gold standard for historic maps we recruited 70+ users from appropriate mailing lists in these domains within four weeks (IFLA DIGLIB and INETBIB lists, and MapHist list, respectively).

Project Background and Organizational Issues

The whole YUMA Multimedia Annotation Suite is an open-sourced evolution of the LEMO annotation tools developed for the European Digital Library (Europeana). We will further develop aspects of YUMA in the following projects:

• SciLink: an EU-funded research project carried out by Bernhard Haslhofer at Cornell University. The focus is on semi-automatic linking and link-maintenance aspects and novel interaction paradigms (e.g., annotations) to support end-users in semantically augmenting Web documents.
• DIALOG: a JISC-funded research project lead by The Open University, UK, where Rainer Simon participates as a project partner. The project investigates how Linked Open Data principles can by applied to express different types of geographical references in textual, visual and tabular documents, focusing in particular on the use case of Ancient World research.

References

1. Chryssoula Boutoura and Evangelos Livieratos. Some fundamentals for the study of the geometry of early maps by comparative methods. e-Perimetron, 1(1):60-70, 2006.
2. G. Braga and S. Gervasoni. Evolution of the Po River: an Example of the Application of Historic Maps. In H. Möoller G.E. Petts and A.L. Roux, editors, Historical Change of Large Alluvial Rivers: Western Europe, pages 113-126, New York, 1989. John Wiley.
3. Bernhard Haslhofer, Wolfgang Jochum, Ross King, Christian Sadilek, and Karin Schellner. The LEMO Annotation Framework: Weaving Multimedia Annotations with the Web. International Journal on Digital Libraries, 10(1):15-32, 2009.
4. Bernhard Haslhofer, Elaheh Momeni Roochi, Manuel Gay, and Rainer Simon. Augmenting europeana content with linked data resources. In Linked Data Triplification Challenge, co-located with I-Semantics 2010, Proceedings of the 6th International Conference on Semantic Systems, pages 40:1-40:3, New York, NY, USA, September 2010. ACM.
5. Yuzuru Isoda, Akihiro Tsukamoto, Yoshihiro Kosaka, Takuya Okumura, Masakazu Sawai, Keiji Yano, Susumu Nakata, and Satoshi Tanaka. Reconstruction of Kyoto of the Edo era based on arts and historical documents: 3D urban model based on historical GIS data. International Journal of Humanities and Arts Computing, 3:21-38, 2010.
6. Alastair W. Pearson. Digitizing and analyzing historical maps to provide new perspectives on the development of the agricultural landscape of England and Wales. e-Perimetron, 1(3):178-193, 2006.
7. C. C. Petit and E. F. Lambin. Impact of data integration technique on historical land-use/land-cover change: Comparing historical maps with remote sensing data in the Belgian Ardennes. Landscape Ecology, 17(2):117-132, 2002.
8. Rainer Simon, Joachim Korb, Christian Sadilek, and Matthias Baldauf. Explorative User Interfaces for Browsing Historical Maps on the Web. e-Perimetron, 5(3):132-143, 2010.
9. David Rumsey and Meredith Williams. Historical Maps in GIS. In A.K. Knowles, editor, Past Time, Past Place: GIS for history, pages 1-18, Redlands, CA, 2002. ESRI Press.
10. Sabine Witschas. Landscape dynamics and historic map series of Saxony in recent spatial research projects. In 21st International Cartographic Conference, number August, pages 59-60, Durban, South Africa, 2003.