CRASH: A Hackathon for Advanced Search in Cultural Heritage

Access to digital cultural heritage data increasingly depends on the ability to search large and heterogeneous collections efficiently. Conventional search methods often reach their limits, particularly when dealing with multimodal collections that combine texts, images and metadata. This is where the CRASH project comes in: as part of a technical hackathon, advanced similarity search methods for cultural heritage data will be tested and further developed. The project aims to bring together experts from the Digital Humanities, machine learning and software engineering to develop prototype solutions for semantic and multimodal search that can be openly reused.  

About the Project  

CRASH is conceived as a technical hackathon that aims to establish the technological foundations for efficient similarity search in Austrian Digital Humanities and cultural heritage contexts. The project focuses on methods that make different types of data, particularly texts, images and metadata, comparable and allow them to be integrated into shared search processes.  

The work addresses several technological and infrastructural questions:  

• Development and testing of prototypes for multimodal similarity search  

• Comparison of client-side and server-side search architectures  

• Benchmarking and optimisation of search algorithms  

• Preparation of openly accessible reference implementations and documentation materials  

Test datasets from Austrian Digital Humanities and cultural heritage repositories will be prepared for the hackathon, including datasets related to GAMS, the Austrian National Library and Kulturpool. In addition, an evaluation framework will be developed to facilitate a systematic comparison of the prototype solutions. The DHInfra.at GPU cluster will be available as the technical infrastructure and can be accessed remotely via VPN. The organising team consists of members of the University of Graz, Graz University of Technology and the Know-Center. The project’s thematic and organisational work packages have already been further developed. The hackathon will follow a focused structure based on advance team formation, clearly defined roles and regular integration stages. A shared online communication channel will also support coordination among the participants before the hackathon begins. The planned outcomes include functional prototypes, an open-source reference implementation and materials for a position paper or demo publication. In the longer term, the approaches developed within the project are intended to be suitable for integration into existing cultural heritage platforms and to provide a foundation for further collaborations and projects.