Digital Access as Archival Reconstitution: Algorithmic Sampling, Visualization, and the Production of Meaning in Large Moving Image Repositories

• Main Authors: Eef Masson, Christian G. Olesen
• Format: Article
• Language: English
• Published: Université de liège 2021-05-01
• Series: Signata
• Subjects: digital; image; categories; objects; interpretation; interaction
• Online Access: http://journals.openedition.org/signata/3011

The article explores how the analysis and visualization of sensory features in digitized moving images contributes towards the reconstitution of audiovisual archives, and how this affects how the objects and collections that make up those archives acquire meaning. In doing so, it takes inspiration from the ongoing research project The Sensory Moving Image Archive (SEMIA). SEMIA was born out of the observation that users, in accessing repositories of digitized moving images, are constrained by current practices of archival description. Institutional catalogues and collection management systems generally make use of so-called ‘semantic’ labels: keywords or other tags that serve to positively identify entities featured in or otherwise associated with specific items (e.g. people, locations, dates, genres, etc.). Those labels are either produced manually (as a result of which they are also highly fragmentary) or, increasingly, with tools for automatic or semi-automatic metadating. However, reliance on semantic descriptors implies a logic of targeted search, which presupposes that users know what they are looking for; moreover, searching is profoundly shaped by the interpretive frameworks that govern labelling. Arguably, this poses restrictions on what users can find or how they can relate archival objects to each other—but ultimately also in terms of how they can use or reinterpret the contents of archives. The SEMIA project team sets out to explore whether, and if so how, visual analysis and visualization of the sensory relations between items can help provide an alternative, affording more exploratory forms of browsing.In this article, we reflect on what this effort entails in terms of how meaning is assigned in the context of moving image archives. What do the different digital transformations that archival objects undergo entail in terms of how they are understood, also in relation to each other? How do the conditions for meaning production get shaped, at different points in the processes of computational analysis and interface design? What are the merits, both of the transformation of images and archives and of the attendant possibilities for new meaning being opened up, in terms of archival access and reuse? In exploring these questions, we work towards a consideration of the role of serendipity in how users encounter, and draw meaning from, database objects, in relation to the tool prototype the SEMIA team is currently developing.