Adaptation of Multimedia eBooks

Ine Langer

Harz University of Applied Studies
Wernigerode, Germany
ilanger@hs-harz.de

Abstract

Multimedia presentations use techniques which also can help making them accessible for people with print disabilities. The different requirements of print disabled readers lead to the notion of rich media documents which contain redundant alternatives of the same information. A personalization process is required to adapt these rich media documents to the needs of an individual reader. Besides this presentational adaptation, it is also necessary to support the reader in building a mental model of the document structure based on temporal and interactive issues related to print disabilities. We describe an implementation of personalizing both multimedia content and document structure.

1   Introduction

Information in various electronic formats covers every imaginable topic and is easily available to everybody. The rapidly growing performance of computer systems and their miniaturization allows for a widespread use of multimedia documents. Most of these multimedia presentations are designed for mainstream recipients , although they use techniques which also can help making them accessible for people with print disabilities. Print disabled user groups are blind, partially sighted, deaf and dyslexic people. Their individual requirements concerning media formats and presentation techniques differ greatly.

Interactive multimedia systems raise other issues on reading strategies and navigation. A multimedia book, as every printed book, can be read in different ways, according to the reader's goals and expectations. Reading for pleasure usually is a linear process (for example reading a novel), whereas the goal of finding a particular piece of information is often achieved by the use of indices or search engines and jumping to different parts of the documents without reading all of them. The integration of both accessibility and navigability in multimedia books is investigated by the MultiReader project whose approach is described in the following section.

2   Personalized Multimedia Books

Creating multimedia books for all users means to fulfil the expectations of mainstream readers as well as the requirements of print disabled users. The following needs can be identified within the different groups of print disabled readers (Petrie et al., 2002). Blind readers prefer text as Braille output and audio presentations, whereas deaf users appreciate graphics, video clips and translations into sign language. Partially sighted people need enlarged views, short lines of text with increased inter-line spacing or special colours and contrast. The latter applies to dyslexic users as well and can be used in combination with animation of text phrases based on highlighting. Table 1 summarizes the user groups and their preferred presentation formats.

Table 1: Summary of User Requirements

Combining the preferred media of all user groups results in rich media documents which contain redundant versions of the same information. Take for example a video clip. It would be accompanied with captions or sign language translations for the Deaf and audio descriptions for the Blind. Of course these alternative representations must be synchronized with the video clip.

The authoring effort of rich media documents is quite high, but this is rewarded by good possibilities of personalization. Further, it allows for equivalent alternatives, that is presentations equally appealing to each type of user. Often this is not the case for documents that have been made accessible at a later stage after the original authoring process.

2.1   One Document for All

The MultiReader system works on the basis of rich media documents and the separation of content from presentation. Figure 1 shows the basic system architecture.

A MultiReader book consists of a main file and several source files which are rich media documents. These source documents are XHTML+SMIL (W3C, 2002) files with special mark-up to identify media elements which are unique to particular user groups. All necessary media objects are provided by the author and delivered with the book. XHTML+SMIL combines the well known Hypertext Mark-up Language (HTML) with the timing and synchronization mechanisms of the Synchronized Multimedia Integration Language (SMIL). XHTML is the XML conformant derivative of HTML which allows for document processing with easily available XML parsers and XSLT engines. Further, XHTML+SMIL documents can be displayed by industry standard web browsers. Therefore a MultiReader book can be used even if the MultiReader system is not available, nevertheless the personalization functions will be missing. The MultiReader system maintains user profiles for different user groups to fine tune enriched documents to the reading requirements of each individual user. User profiles are stored in XML files.

In an XSLT process the source documents are evaluated against the user profile, and undesired media elements are removed. The resulting document nodes then are arranged in a particular reading order which is specified by a so called tour. This concept is detailed in the next section.

Visual layout of the book is controlled by means of Cascading Style Sheets (CSS), stored in separate files. The user profile contains information about which style to use.

2.2   All Documents for One

Once the user has specified a suitable profile, he/she should be able to follow an appropriate reading strategy to get the desired information from the book. Different navigation aids, such as table of contents, index, search, site map, history etc. are likely to be used while reading (Conklin, 1987). Another navigational concept is the tour. By providing a linear view of the data it prevents readers from getting lost in large or unstructured document sets. Tours, as sequences of hypertext nodes, can be found on some web sites (CAST, 1999), and even in existing electronic book formats (OeB Forum, 2002). Both these implementations support different tours around one set of documents (provided by the author), and even user defined tours.

In the MultiReader system tour information is stored in an XML file. As opposed to the above examples, a MultiReader tour may contain sub-tours. This reflects the hierarchical structure of chapters and sections in traditional books. For example, a tourist guide may describe a trip around a number of cities and one or more sightseeing walks for each of them. Using such a structured approach, it is easier for the reader to skip details and continue with the next top level element, while still following a predefined path.

The author of a multimedia book is expected to provide at least one tour as the default reading order for the user to start with. Then the user is free to compile his/her own tour definitions.
The creation of user defined tours requires presentation and interaction techniques appropriate to the user's profile. This may include both static and time dependent media, for example written text, interactive maps, sign language and speech. No matter which media are chosen, an overview over the available tour objects (items which are to be visited) must be provided. In the tourist guide example a city map could inform the user about sights that can be visited, their importance and their geographical location. This information may be not as obvious from textual representations. In order to make such relations clear to all users, it is necessary to describe their semantics in a universal format and then present them in an appropriate way.

Figure 2 shows a first approach based on interactive maps which has been implemented in the MultiReader system. Parts of the city map of London are represented as Scaleable Vector Graphics (W3C, 2001) which allows for magnification without loss of image quality. The user selects tour points by clicking on the corresponding elements in the map. When the user acknowledges that the list is finished, a tour file will be stored and is then ready to be applied to the document set.

3   Conclusions and Outlook

Enriched multimedia documents contain more than multiple versions of the contents in order to address users with print disabilities. In order to facilitate the navigation of the document both structure and temporal constraints have to be reflected in the mark-up of the document. XML-based mark-up allows an efficient implement of the personalization processor.

So far the adaptation of both the multimedia contents and the tours takes place upon user request. Evaluation of the MultiReader system with visually impaired users is planned and pilot testing shows a high level of acceptance. Multiple interaction techniques for users requiring a non-graphical tour selection mechanism is under preparation.

A further enhancement would be an adaptive system which adjusts presentational settings automatically. This could affect the presented media elements as well as the tour objects. In a tourist guide for example, the time that the reader spends with certain pages could be measured and used as a cue to which kind of information he/she is interested in. As a result irrelevant items could be removed from the tour. When making such a decision, it is important to consider the presentation format. A user working with time dependent media (video clips, spoken text) would spend more time than someone who quickly scans text visually. Fine-grained temporal modelling will be required in order to understand user’s interactive behaviour when navigating multimedia documents.

4   Acknowledgements

The MultiReader Project is funded under the IST Programme by the Commission of the European Communities (Project IST-2000-27513). The Consortium consists of the City University London (United Kingdom), the Electrotechnical Engineering Department (ESAT) of the Katholieke Universiteit Leuven (Belgium), the Royal National Institute for the Blind (United Kingdom), the Federation of Dutch Libraries for the Blind (the Netherlands), the Harz University of Applied Studies (Germany) and Pragma (The Netherlands). The construction of a graphical tour selection mechanism became possible through Daniela Mack’s help.

References

CAST (1999). Guided Tours. Retrieved February 11, 2003, from http://www.cast.org

Conklin, J. (1987). Hypertext: An Introduction and Survey. IEEE Computer, 20 (9), 17-41.

Open eBook Forum (2002). Open eBook Publication Structure Specification Version 1.2. Retrieved February 10, 2003, from http://www.openebook.org

Petrie, H., Langer, I., Weber, G., Fisher, W., Gladstone, K., Rundle, C., Pyfers, L. (2002). Universal Interfaces to Multimedia Documents. International Conference on Multimodal User Interfaces, IEEE, (Pittsburg, 14-18.Okt. 2002), 319-324.

W3C (2001). Scalable Vector Graphics (SVG) 1.0 Specification. Retrieved February 10, 2003, from http://www.w3.org/TR/SVG/

W3C (2002). XHTML+SMIL Profile. Retrieved February 10, 2003, from http://www.w3.org/TR/XHTMLplusSMIL/