ArtBase User Guide
For a brief introduction to the ArtBase, see our FAQ.
This guide is intended to help you navigate the different ArtBase user interfaces, giving you access to artworks’ metadata and related media files, and querying the data to create data visualizations. Unlike the relational database approach of most traditional collection management systems, Rhizome’s archive of net art uses Linked Open Data (LOD). This is a set of standards for structured, machine-readable and interoperable data on the web. We are using LOD within a Wikibase installation to structure and organize the heterogeneous data needed to describe born-digital artworks.
Browsing the archive
Currently, all artworks in the archive can be browsed by date or by artist’s (or collective’s) name.
A conventional keyword search facility will also be released soon.
See the section “Writing your own SPARQL queries” further below to learn how to get started browsing the ArtBase via a dedicated SPARQL endpoint, or see some examples.
Artworks and variants
Net art challenges the notion of artworks as single, self-contained objects. To be performed and experienced, net art depends upon alignments between hardware and software environments, network protocols, and user interactions. Therefore, a work of net art can evolve over time into various instantiations—either because of intentional actions by the creator(s) of the work, intervention by institutional staff working to preserve or exhibit the work, or due to structural changes in the software or network components of the work. Rhizome refers to these multiple instantiations as “variants.” A key component of the new ArtBase infrastructure is the potential to link a work to all of its variants.
Initially, artists could submit works for accession into the ArtBase as either “cloned” or “linked” objects. In discussions with artists, Rhizome Founder Mark Tribe developed a hybrid model of the ArtBase, where cloned works were stored as file and/or server copies on Rhizome infrastructure, and linked works were merely described in ArtBase and remained under the stewardship of the artist. In the new LOD structure of the archive, a file and/or server copy of a work is referred to as an ‘ArtBase variant’, while links to works maintained by the artist (or a third party) are referred to as an “Outside Link.” As new preservation tools and approaches were developed at Rhizome over time, additional variant types, such as “web archive” or “emulated instance,” became a part of the archive, too. See also Historical background.
Access points and iconography
Artwork variants are directly accessible through dedicated access points available on the main page for each artwork in the archive, e.g. a page like this one. The access points communicate the custody of the variant: is it under the care of the artist or another party (Outside Link) or the archive (ArtBase variant)? In addition, visual cues in the form of icons differentiate between different types of variants. To see more detailed metadata about each variant, you can expand the “Metadata” field. Soon, we are also planning to provide visual indicators about the state of each variant access point, e.g., is it completely inaccessible, partly damaged, or generally functional.
Artwork descriptions
In addition to image representations and access points to different variants, each artwork page may feature a range of descriptive information. During the open stage of the ArtBase accession policy, artists submitted long-form text descriptions and/or artist statements alongside their works. These descriptions are preserved as non-machine-readable, natural language texts and listed within an expandable field on artwork pages titled “Description.” In some cases, short summary texts created by Rhizome staff to introduce the works are also listed within this field. All descriptions are followed by source metadata. “Attribution” denotes who contributed the description, and “inception” denotes the date the text was submitted to the archive.
In some cases, we also present old tags as part of the descriptive data about artworks. Tags were a legacy feature of previous versions of the ArtBase and were crowdsourced among users and Rhizome staff. While tags are no longer an active, searchable component of the database, they hold valuable historical information regarding the types of terms and categorizations popular during various phases of the development of the archive and the net art field as a whole.
Metadata
Next to the expandable “Description” field on the main pages for artworks, there is another expandable field, which holds the complete set of metadata available for a given artwork in the ArtBase. In addition to capturing core bibliographic data for each artwork, the variant data model makes it possible to document a range of data applicable only to a specific variant.
Such data is accessible in a secondary expandable field, and may include inception, i.e. the date the variant was created (as well as the date it stopped being active); type and date of accession into the archive; and further—changes in technical properties and user-interaction requirements.
It also provides space to express the reasons for changes to the variant which may be due to an active intervention by the artist, or caused by a contextual event outside their control, such as a component becoming obsolete. Preservation activities which result in the generation of new variants (for example, the addition of a web archive or an emulation instance) can be captured alongside information about the agent(s) who carried out the action. Variants generated through such actions are associated with the archival institution, rather than the artist.
![A screenshot of the Metadata section for the web archive link variant of untitled[scrollbars], an artwork in the ArtBase](https://artbase.rhizome.org/w/images/2/24/variant_metadata_v2.png)
Writing your own SPARQL queries
Linked Open Data (LOD) databases are not well-suited to traditional faceted search (i.e. search by a limited set of categories or topics), which tends to flatten the complexity of a heterogeneous, networked archive. Instead, the best way to browse an LOD database is to use SRARQL✨.
The LOD infrastructure of the ArtBase enables users to ask new questions from the archival data—questions that go beyond keyword search, or search, based on faceted categories. This is facilitated through a SPARQL endpoint with a graphical user interface (See Browsing the archive above). We refer to Rhizome’s customized SPARQL endpoint as the ArtBase query service. It allows us to query and visualize the data in real-time through a variety of visualization approaches, such as tables, image grids, charts, maps, and more.
In this user guide, we’ll show you how to get started using the query service. You do not need any previous knowledge of LOD or SPARQL to work through this tutorial. We will be using the visual “query helper” interface first, and then gradually adding complexity through manual editing. To get started, you can read the brief introduction to the LOD syntax and the tools available in the query service below, or jump straight to the step-by-step tutorial.
LOD syntax
It is helpful to start with a brief introduction to how linked data in the ArtBase is structured. The core syntax of the data model follows the Resource Description Framework (RDF)—an LOD standard—and is organized in ‘triple’ statements, consisting of: 1) subject, 2) predicate, and 3) object. In Wikibase and the ArtBase query service, these LOD terms are referred to as 1) item, 2) property and 3) value. Items are the main subjects of a particular data statement. Properties connect these items to related data values. For example, a specific artwork (the item) is connected to the name of its creator (the value) via the property “artist”. Understanding this core relationship between data points in the archive will help you understand how queries are constructed later in this tutorial.
A guide to tools and examples in the query service
The query service interface is split into several main areas. The top bar provides navigation links and access to pre-composed example queries. Below it, there is a series of tools in a sidebar to the left. The top half of the page is split in two areas: the visual composing area to the left (what is referred to as the “query helper” interface in this guide), and the manual composing are to the right, where you can manually write SPARQL expressions. The bottom half of the page is the results area, which is populated with results once you “run” a query.
The query service sidebar includes the following tools:
1. Opens and closes the “query helper” interface.
2. Opens the query window in full screen mode.
3. Opens a list of prefixes from different SPARQL endpoints that can be used in the query service (helpful for federation).
4. Formats the query for readability (a “beautify” tool).
5. Opens the examples folder. (Can be accessed also from the top navigation bar in the query service)
6. Retrieves a previous query.
7. Clears the query.
8. Builds a URL to this query.
9. Run your query.
To help you get started with ideas what is possible to query for, we have prepared various examples which can be accessed from the #5 icon shown above and from the top navigation bar.
The examples menu offers several options to preview results or see the SPARQL code of the queries. Examples are split into sections. Simple queries provide general use-cases for the query service. The "Template queries" are annotated and can be used as models to easily change a single parameter and get other results. Note that to modify most example queries you will need to modify the SPARQL code in the window to the right of the query helper interface. We also list examples that highlight the use of different types of data visualizations. Lastly, federated query examples demonstrate the possibilities to draw data from across multiple LOD sources, including Wikidata.
Once you run one of the example queries, or your own, you will see the following set of additional tools and information panels appear below the composing areas:
1. Let’s you choose among a range of different data visualization types. Not all of them area available for all queries. For example, unless the query looks for images, the image grid visualization will not be highlighted as a possible option to choose.
2. Opens a help page about data visualization functionalities. Note: This will take you to a generic Wikimedia page which is outside Rhizome’s infrastructure and not all the information is applicable to the ArtBase.
3. Updates the results count every time you run your query.
4. Provides code snippets for your query which can be used in various other code environments.
5. Let’s you download your query results in a variety of standard formats.
6. Let’s you share your query results.
Step-by-step tutorial
To learn how to compose your own query, follow the step-by-step tutorial. The example query we use in the tutorial shows us all artist collectives in the ArtBase and who their members are.
Further reading on LOD and the ArtBase data model
So far we this user guide has only introduced a cursory view of how structured data is represented in Wikibase. Statements are composed of properties associated with items and their respective values. Statements can have references, too. Without a reference, a statement is simply a claim. Claims can have qualifiers—these are sub-properties which add additional detail about a claim—e.g. what time period a claim relates to, or what is it’s source. Adding qualifiers to claims enriches the data set and can create more interesting and nuanced results in data queries.
The unique advantage of Wikibase over other knowledge management systems is that there are no pre-set hierarchies or ontologies. Wikibase can function as an ontological sandbox and space for experimentation—there is no need to follow prescribed standards or conventions utilised by other organizations. Any items can be created within the database, as well as any properties, though properties should remain a limited set, in order to make querying easier. The ontology can change and be updated as needed over time. To get a quick overview:
- See a list of the common properties and value examples currently in use in the ArtBase.
- See an extended data model proposal, which is currently only partially implemented.
Lastly, the embedded support for standard RDF data description and SPARQL queries in Wikibase enables linking the ArtBase to other structured linked data databases. This facilitates less redundancy in data description—for example, if an artwork, a person, a place or piece of software relevant to the ArtBase is described in another LOD database, it can be linked to Rhizome’s Wikibase and vice versa. The resulting links enable federated queries—queries which can check for information available in multiple databases, and display results pulled from the respective sources in a single interface (e.g. the ArtBase query service, or another SPARQL endpoint). Such data may include artist’s biographical data (see this example), artwork exhibition data, or data about specific software items, needed for the preservation of an artwork in the archive.
Further reading:
Blaney, J. (2017) "Introduction to the Principles of Linked Open Data," The Programming Historian 6. Available here.
Fauconnier, S., et al. (2018) "Many faces of Wikibase: Rhizome’s archive of born-digital art and digital preservation – interview with Dragan Espenschied, Lyndsey Moulds, Lozana Rossenova," Wikimedia blog, 6 Sept 2018. Available here.
Jones, E., and Seikel, M. (eds.) (2016) Linked Data for Cultural Heritage, An ALCTS Monograph. Chicago: ALA Editions, an imprint of the American Library Association.
Rossenova, L. (2020) "ArtBase Archive—Context and History: Discovery Phase and User Research 2017–2019." Available here.
Rossenova, L., et al. (2019) "Provenance for Internet Art Using the W3c Prov Data Model," in Proceedings of iPRES 2019, 16th International Conference on Digital Preservation, Amsterdam, The Netherlands. pp. 297-305. Available here.