- The paper introduces the OPALES project as a framework for indexing, annotating, and exploiting extensive digital audiovisual archives in humanities.
- The paper details a client-server architecture with specialized tools like a video explorer and ontology builder that enhance multimedia research workflows.
- The paper identifies challenges in user adaptability and proposes simplified interfaces, pre-defined models, and improved standards for future archive systems.
This paper "Digital audiovisual archives in humanities" (2503.13452) by Peter Stockinger describes the audiovisual archive program initiated by the Maison des Sciences de l'Homme (MSH) in Paris in 2001 and the associated OPALES project, which aimed to develop tools for indexing, annotating, and exploiting this archive. The core problem addressed is how to make a large collection of audiovisual material in the humanities (social and human sciences) effectively usable for researchers, teachers, and professional users.
The MSH audiovisual archive program had several goals:
- Production of scientific knowledge: Creating a content base for new knowledge and shared standards, and serving as an observatory for research trends.
- Conservation: Preserving scientific contributions and creating an open archive of scientific heritage.
- Diffusion: Providing central access to dispersed knowledge and creating reusable elements for multi-support publishing.
- Task-oriented exploitation: Building a sharable knowledge base for research/teaching communities and professionals, and creating a "competence pool" for young researchers.
The content of the archive primarily consisted of footage from research seminars, symposia, workshops, lab life, specific competencies, and in-depth interviews with researchers. As of August 2003, the archive had 450 hours of scientific video online, with more to be digitized. Access was provided via the MSH website and a dedicated media portal. The principal information unit is the "event" (interview, seminar, etc.), which is represented as a dynamic website based on data in a relational database and indexed in the media portal. The website for an event typically included a home page, a video page for consulting the material, and a thematic index page linked to concepts and their appearance in videos. The canonical production chain for an event involved shooting, digitizing, cutting footage into files, post-editing, thematic description and indexing, exporting to multiple formats (asf, mpeg), registering data in a database, generating a dynamic website, and broadcasting it within the MSH archive.
The paper outlines the needs of the main users (researchers, teachers, professionals):
- Creating personalized work spaces for specific usages.
- Identifying and selecting relevant audiovisual and textual segments.
- Specifying personal viewpoints for classifying and describing selected segments.
- Creating "personalized" ontologies, thematic grids, and annotation categories.
- Sharing and collectively updating viewpoints and ontologies.
- Creating personal or group-specific archives.
- Researching and exploring indexed and described segments based on specific viewpoints.
- Producing new "montages" (hyper-documents) from selected segments.
- Editing these montages for various formats (website, CD-ROM, DVD, etc.).
These needs were the basis for the French OPALES project (Outils pour des Portails Audiovisuels Educatif et Scientifique), an R&DT initiative focused on developing an integrated environment prototype for describing and producing new montages from audiovisual and textual segments based on specific viewpoints.
The OPALES environment prototype, developed between 1999 and 2003, had a client-server architecture comprising the Opales client, a video server, and an environment server. The Opales client interface included a research/exploration interface and a working interface. The working interface featured several specialized tools:
- Video explorer: An enhanced media player for selecting segments and zooming on frames.
- Ontology builder: For creating hierarchies of themes (concepts) and relations representing a domain's knowledge.
- Conceptual graph editor: For building configurations (graphs) of selected themes and relations to represent specific scenes within segments.
- Point of view editor: A generic formulary allowing users to define custom features for describing segments based on their specific interests (e.g., rhetorical nature, credibility, importance).
- Video segment indexing and annotation interface: For attaching themes, conceptual graphs, viewpoints, or free annotations to segments or parts of segments.
- Segment montage editor: For creating oriented navigation paths (hyper-documents) through described segments.
- Work space tool: For managing personal or collective work spaces and access permissions for resources (ontologies, descriptions, etc.).
The paper presents a typical user scenario involving a researcher building a personal library on a topic like industrialization. The steps involve:
- Creating a personal work space.
- Exploring the MSH archive and creating a base library of relevant events ("annotated bookmarks").
- Identifying and selecting specific audiovisual/textual segments from events into a personal segment library using the video explorer.
- Specifying an ontology for their research domain, potentially creating a hierarchy of themes (notional, rhetorical, contextual) and relations (classifications, practical/epistemic inferences, modalisation, localizations) using the ontology builder. Then, indexing segments by attaching themes to specific temporal positions.
- Defining typical scenes within segments using the conceptual graph editor, importing themes/relations from the ontology, building conceptual graphs representing scene types, and attaching these graphs to corresponding temporal positions in segments, possibly adding keywords or textual descriptions as referents.
- Customizing the description model using the point of view editor to include specific, personalized features relevant to the researcher's analysis.
- Producing a complex hyper-document (montage) from the selected and described segments using the montage editor.
- Managing sharing permissions for their resources (ontologies, graphs, viewpoints, descriptions) using the work space tool.
The project identified significant difficulties and limitations. A major challenge was the users' (researchers', teachers') working habits and lack of preparedness to make explicit their practical classification and description schemas, which is necessary for working with ontologies and viewpoints. Eliciting these tacit schemas requires methodological knowledge often not universally possessed within research communities. This complexity also meant the tools felt too "rich" and "general" for the average user.
Future directions proposed for a second phase of the OPALES project included:
- Providing simplified and standard working interfaces tailored to specific task profiles (e.g., "realize a course," "build a personal archive").
- Providing pre-defined models (ontologies, typical scenes, viewpoints) within the editors.
- Transforming the tools into web-based services.
- Introducing genuine multilingual support.
- Enhancing specific tools, like the montage editor.
- Developing genuine multi-support publishing capabilities.
- Integrating principal metadata standards (DC, OAI, MPEG 7, TVA).
The paper concludes by highlighting that while OPALES was crucial for enabling specific user groups to exploit and enrich audiovisual archives, significant effort was still needed to overcome user-related and technical challenges and make the tools more accessible and standards-compliant.