Temenushka Ignatova

Towards a Digital Document Archive for Historical Handwritten Music Scores

Contemporary digital libraries and archives of music scores focus mainly on providing efficient storage and access methods for their data. However, digital archives of historical music scores can enable musicologists not only to easily store and access research material, but also to derive new knowledge from existing data. In this paper we present the first steps in building a digital archive of historical music scores from the 17th and 18th century. Along with the architectural and accessibility aspects of the system, we describe an integrated approach for classification and identification of the scribes of music scores.

Model-Driven Development of Content-Based Image Retrieval Systems

Generic systems for content-based image retrieval (CBIR), such as QB1C [7] cannot be used to solve domain-specific retrieval problems, as for example, the identification of manuscript writers based in the visual characteristics of their handwriting. Domain-specific CBIR systems currently have to be implemented bottom up, i.e. almost from scratch, each time a new domain-specific solution is sought. Driven by the recognition that CBIR systems, although developed for different domain-problems, have similar building blocks and architecture, we propose a framework model for the development of CBIR systems. The aim of the framework model is to support the design of domain- specific CBIR-systems on a conceptual level by reusing architectural design and building blocks. The model-driven development of CBIR systems can thus be realized by software components for modeling the data structures and the update, storage and retrieval operations using the predefined framework model, and by components for mapping this model onto a specific implementation platform.

Utilizing a multimedia UML framework for an image database application

To support the design of data models for multimedia applications, we employ the concept of a framework introduced in object-oriented design. We define a UML framework, which can be used for deriving application-specific multimedia database models. With the UML framework, we define the core elements of a multimedia database model, such as mediatype- and application-independent structure, content, relationships and operations. Thereby, the advantages of using UML for representing multimedia data as well as shortcomings of this approach are discussed. Furthermore, we describe the utilization of the UML framework for the instantiation of a model for an image database of scanned handwritten music scores.

Knowledge-Based Scribe Recognition in Historical Music Archives

For the content-based management and access to domain-specific data in digital libraries, special domain-knowledge and knowledge processing functionality are required. However, the integration of knowledge components has not yet become an integral part of existing digital library systems. The current paper represents the realization of a digital archive of historical music scores, integrating special domain-specific data and functionality for writer identification in historical music scores. We introduce the basic formalisms and heuristics for the representation of handwriting characteristics. To compare two handwritings we propose the usage of a normalized, weighted Hamming distance function to calculate the degree of similarity between their handwriting characteristics. For the identification of writers we employ the k-nearest neighbor method to build clusters of similar writers, based on the calculated distance. And finally, we represent and evaluate the test results from the prototype implementation of the system.

Identifying Semantic Constructs in Web Documents to Improve Web Site Accessibility

Misleading user interfaces and overloaded web sites are some of the reasons why users avoid certain web sites while searching for information on the world wide web. In order to improve the usability and accessibility of web sites, techniques which take the semantic structure of the web documents into account have to be employed. The semantic analysis approach described in this paper aims at recognizing those parts within web documents that are particularly relevant for a specific usage scenario. Different combinations of syntactical constructs are mapped to different analysis classes whose semantic constructs correspond to concrete interaction tasks predefined in task models. The task models are used as a workflow guiding the user through the shopping procedure, whereby for each identified task a corresponding semantic concept on the web site is identified. The described techniques will be embedded in a so called screen reader application for visually impaired people who do not have the ability to use a graphical display.