A NEW ARCHITECTURE

What you see here is a result of several years of dedicated work by Jeff Wallman and Chris Tomlinson, the developers of the digital library management system (DLMS) and key subject matter and technical advisors Sandy Roberts, Michele Martin and Derek Kolleeny.

Tibetan display. An important feature is the ability to display work titles, personal names, places, and keywords in both Wylie Romanization and Tibetan script. Users must have installed Tibetan machine web font, a public domain font, which can be downloaded from the site and download into the users local "font" directory. Visit our help page on displaying Tibetan

The Tibetan display works well on all major browsers with the exception of Microsoft Internet Explorer browser where a bug in the browser causes sporadic extra spacing next to vowels. For optimal results, we suggest using a freely available public domain browser such as Netscape (Mozilla), Firefox, or Opera. There are still problems in the alignment of the Tibetan script with the roman letters. Hopefully with the distribution of Unicode Tibetan these will look better.

Design Process. The design team developed the website and underlying database with a variety freely-available public domain tools – the eXist XML database, an XQuery/Apache-Cocoon engine for processing data to the web and an Eclipse-based user interface for data management. At the outset of the project, a considerable amount of time was dedicated to the development of an accurate, flexible and extensible UML model of all the objects in the system. Using this UML model, an XML schema was created which became the basis for an EMF generated model layer of the DLMS. The benefit of this approach is that any changes, modifications or additions to the conceptual elements of the system can be easily propagated to the user interface and then to the website. This essentially means that because data model and associated layers of software are inherently flexible and extensible, the DLMS system will last well into the future.

Initially, the team used the public domain software Fedora as the repository framework. However, a key feature of the TBRC project, the cross-referencing resource index, was not available in Fedora during the DLMS development cycle. Therefore, the team built the DLMS system to be forward compatible with Fedora so that when that index feature is installed in Fedora, a simple cut-over strategy can be employed to allow the DLMS system to use Fedora.

Knowledge Base. All of the current data was migrated into the new DLMS software. The resulting Knowledge Base has over 30,000 Works, 2,500 subjects, 10,000 persons, 5,000 places and over 100 monastic institutions, aristocratic families and clans. New models have been created to allow for additional objects to be entered - detailed outlines of large collections, lineage transmissions and a subject heading taxonomy. The DLMS also has a production manager module for scanning and quality assurance to support the digital archiving.

DIGITAL LIBRARY MANAGEMENT SYSTEM (DLMS)

The resulting software is a powerful delivery system for digitalized Tibetan materials and Knoweledge Base documents. The key technologies and features of DLMS are:

• Image Repository. TBRC holds an archive of high resolution archival TIFFs. Web-access to the archival TIFFs is managed by an image server that generates on-the-fly JPEGs for viewing from the archival TIFFS. There is a zoom feature to allow accurate analysis of Tibetan characters. The underlying software that generates the JPEGs is based on the Java Advanced Imaging API.
• PDF creator. In addition to web delivery, the DLMS system produces PDF on-the-fly. This allows licensed users to create PDFs for download for their own personal use. The PDF converter is based on the PDFLib API.
• XML. DLMS is an XML-centric software system. Based on the eXist, native XML database, all objects are modeled as XML documents. DLMS delivers these documents to the web via XQuery web application that transforms the XML using XSL Apache Cocoon.
• Apache. DLMS uses a variety of server side technologies to deliver Web content including Apache Tomcat and Apache HTTPD Server.
• Portability. DLMS was intended to reside in many places - the webserver hosting tbrc.org, our partner scanning organization in New Delhi, India but also on laptops and remotely installed digital libraries all over the world. Portability is intended to provide access to the digital library, where an internet connect is unavailable or unreliable.

Hardware Environment The TBRC Data Center is managed directly by TBRC technical staff. The Center is connected directly to the Internet via a a multi-homed 100Mbps connection at a leased facility providing 24X7 access at 470 Vanderbilt Avenue, Brooklyn, NY. The server is a high performance dual Intel 3.6GHz (64-bit) processor machine with 4 GB RAM, dual power supply and six 300 GB 10K SCSI hard drives (1.8 TB), running RedHat Linux (Fedora Core 5).