Information Preservation in ARIADNE

Preserving digital information is a necessary commitment to the future. In ARIADNE, a project of the Digital Publishing Group at ICAT, we are developing an integrated information system for news processing and publishing. In this paper, we present our perspectives on various information preservation issues addressed by this research. These include the semantic preservation of information classification schemes, preservation of the layout of dynamically generated documents, preservation of the linkage to external collections, and the economic sustainability of the news archive. INTRODUCTION Preserving or archiving digital information is, today, an important and necessary commitment to the future. How will we make available today’s news to future readers? Giving the importance of this topic, the Digital Publishing Group of ICAT is studying new methods for preserving and processing heterogeneous information in organizations, combining multiple databases under a common framework. In project ARIADNE, jointly developed with Público, a national daily newspaper, we are building a new digital publishing structure, where all the information used and produced by journalists is organized in a common database (containing both data and metadata for collections maintained outside the organization). From the information in this digital library, we generate publications in digital format. Público already maintains an archive of all the editions of its paper publications. This is a profitable unit within the company. The archive is used by the newspaper journalists and provides services to external entities. However, as we move into on-line publishing (we are about to release several new publications which will be available exclusively on-line) there is a need to define the processes for archiving and retrieving previously published on-line information. These new on-line publications differ significantly from the previous generation, where we were doing little more than creating on-line replicas of the paper editions. Our new publications are beginning to behave more and more as interactive user interfaces to databases of multimedia presentations. In the next section we present the global architecture of our system, and then proceed with a more detailed discussion of the preservation issues in our digital library. ARCHITECTURE The architecture of ARIADNE is based on large multimedia data repository, which holds various collections of documents, newspaper articles, databases of readers and authors, places and events. For some collections, namely external publications, we only keep metadata and links for the articles. The global architecture and the main information flow are shown in Figure 1. Several sources, including news agencies feeds, articles created for the paper edition of Público and external publications provide news items to the ARIADNE repository. Each article received is submitted to a preprocessing stage, where its metadata is extracted. The articles and news feeds are then converted into a common format based on XML[9], and archived in the collection repository with the software module Loader. Editions of electronic publications are built on a second stage, using another module, Generator, which selects a group of articles archived in the collections repository and packs them into presentations (or editions). This process finishes by converting the XML sources to HTML, making articles viewable from the current generation of web browsers. With this strategic approach we intend to overcome possible changes in data format standards and sustain our archival mission[3]. Collections Repository Editions (HTML) Publication Templates External Search Engines Internal Search Engines Público Paper Version News Agencies Feeds Loader Generator Article (XML) External Publications