ARTLAS: Mapping Art-Technology Institutions via Conceptual Axes, Text Embeddings, and Unsupervised Clustering
The global landscape of art-technology institutions has seen remarkable growth and diversification in recent years. This includes a variety of entities such as festivals, biennials, research labs, conferences, and hybrid organizations. However, there remains a significant gap in systematic frameworks for analyzing the multidimensional characteristics of these institutions. To address this issue, a new paper proposes ARTLAS, a computational methodology developed to map 78 cultural-technology institutions into a unified analytical space.
Overview of the ARTLAS Methodology
ARTLAS employs an eight-axis conceptual framework that includes:
- Curatorial Philosophy
- Territorial Relation
- Knowledge Production Mode
- Institutional Genealogy
- Temporal Orientation
- Ecosystem Function
- Audience Relation
- Disciplinary Positioning
This framework is complemented by a text-embedding and clustering pipeline that characterizes each institution through qualitative descriptions along these eight axes. The qualitative descriptions are encoded using E5-large-v2 sentence embeddings, which are then quantized into TF-IDF feature vectors. Dimensionality reduction is achieved through UMAP, followed by agglomerative clustering with an average linkage method, resulting in a composite score of 0.825, a silhouette coefficient of 0.803, and a Calinski-Harabasz index of 11,196.
Data Analysis and Findings
To further understand the institutional landscape, non-negative matrix factorization is utilized to extract ten latent topics. Additionally, a neighbor-cluster entropy measure is employed to identify boundary institutions that bridge multiple thematic communities. The findings reveal coherent groupings among the mapped institutions, which are categorized into distinct clusters:
- Art-Science Hub Cluster: Anchored by prominent institutions such as ZKM and ArtScience Museum.
- Innovation and Industry Cluster: Comprising organizations like Ars Electronica, transmediale, and Sonar.
- ACM Academic Community Cluster: Including conferences such as TEI, DIS, and NIME.
- Electronic Music and Media Cluster: Featuring festivals like CTM Festival, MUTEK, and Sonic Acts.
Interactive Visualization Tool
To enhance stakeholder engagement, an interactive web-based visualization tool has been developed using React. This tool allows users to explore institutional similarities, thematic profiles, and cross-disciplinary connections, thereby facilitating a deeper understanding of the diverse landscape of art-technology institutions.
Conclusion
The ARTLAS project represents a significant advancement in the field of institutional ecology within the cultural-technology sector. By combining a replicable, data-driven approach with comprehensive qualitative analysis, ARTLAS offers valuable insights into the complex interrelations among cultural-technology institutions. This methodology not only enriches academic discourse but also serves as a practical resource for stakeholders seeking to navigate the evolving landscape of art and technology.