How to teach digital library data to swim into research

Christoph Schindler, Cornelia Veja, Marc Rittberger, Denny VrandecicI-semantics 2011 · DOI

TL;DR

This paper proposes a novel approach to bridge the gap between digital libraries and virtual research environments (VREs) by leveraging Semantic Web technologies. The authors demonstrate how integrating bibliographic data from libraries into a VRE built on Semantic MediaWiki can empower researchers with flexible tools for data exploration, annotation, and analysis, ultimately making digital library resources more actionable for research.

What Problem Does This Paper Address?

The paper tackles the challenge of interoperability between digital libraries and virtual research environments (VREs). While digital libraries have made vast collections of digitized cultural artifacts and research data accessible, researchers often struggle to effectively integrate and utilize this data within their VREs. This disconnect stems from differing practices, tools, and data structures between library and research communities. The authors highlight that VREs, while aiming to enhance research, often face interoperability issues and a lack of lightweight tools. The "complexity deluge" in the humanities, characterized by fuzzy, incomplete, and distributed data, further exacerbates this problem, pushing interoperability beyond simple standardized protocols. The paper aims to provide a solution that aligns research practices with library resources by making digital library data more "swim-able" into research workflows.

Methods

The core of the paper's methodology lies in the Semantic MediaWiki for Collaborative Corpora Analysis (SMW-CorA) project. This project integrates digital library data into a VRE using a three-pillar approach:

  1. MediaWiki as a web-based repository: Leveraging MediaWiki's collaborative platform for storing and organizing digital resources.
  2. Semantic MediaWiki (SMW) annotation mechanisms: Utilizing SMW's ability to add semantic annotations to wiki pages, enabling data interlinking and schema rearrangement. This allows for richer metadata and more sophisticated querying.
  3. Semi-automatic workflow management: Developing a flexible workflow for data integration, including metadata harvesting, transformation, and upload. This process is driven by XML configuration files and XSLT transformations, making it adaptable to different data sources.

The project also employs a participatory design approach, actively involving researchers in the design process to ensure the VRE meets their specific needs. This includes iterative development, agile computing, and gathering feedback through talks, site visits, group discussions, and rapid prototyping.

Datasets

The paper uses bibliographic data from the Scripta Paedagogica Online (SPO), a digital library hosted by the Library for the History of Education (BBF) at DIPF. This corpus contains approximately 22,000 articles from over 20 lexica, with bibliographic data stored in the Allegro-C database. The integration process focuses on transforming and importing this bibliographic metadata into the SMW-CorA VRE.

Key Insights and Ideas

  • Bridging the "Complexity Deluge": The paper argues that Semantic Web technologies, when integrated with VREs, can effectively address the "complexity deluge" in the humanities by providing flexible ways to represent and interact with heterogeneous and interconnected research data.
  • "Teaching Data to Swim": The central metaphor of "teaching digital library data to swim into research" effectively captures the goal of making library resources more fluid and actionable within research workflows, rather than static objects to be passively consumed.
  • Lightweight Ontology for Mapping: The development of a "light-weight mapping ontology" is crucial. This ontology defines how library metadata entities (like lexicon, volume, article) are mapped to MediaWiki pages and SMW properties, enabling a structured yet flexible integration.
  • Participatory Design is Key: The success of the SMW-CorA project hinges on its participatory design approach, ensuring that the VRE is not an isolated application but is closely aligned with the actual research practices and needs of its users.
  • Beyond Simple Interoperability: The paper emphasizes that true interoperability requires more than just syntactic matching; it involves semantic understanding and the ability for researchers to dynamically re-arrange and create new relationships between data.

Key Results

The paper details the successful implementation of a semi-automatic data integration process from the SPO digital library into the SMW-CorA VRE. Key achievements include:

  • Automatic metadata integration: The system can automatically integrate metadata from digital libraries into SMW.
  • Extensible integration process: The approach is designed to be extensible to heterogeneous data integration processes by requiring only an attached workflow template and XSLT transformation.
  • Layered approach to interoperability: The system offers a layered approach to achieving interoperability at both syntactic and semantic levels with a low technological entry barrier.
  • Support for research workflows: The VRE provides functionalities that support researchers in their initial research steps, such as annotating and categorizing articles, and facilitating collaborative discussions.

Limitations and Future Work

The authors acknowledge several limitations and outline future work:

  • Interlinking annotation and analysis levels: A key open aspect is the interlinking of different levels of annotation and analysis within the research project, ensuring adequate granularity and flexibility.
  • Alignment with other disciplines: The SMW-CorA VRE needs to be aligned with research practices from other disciplines to broaden its applicability.
  • Full research life cycle integration: The current integration process covers only the initial steps of the data life cycle (create, update, publish). Future work will focus on addressing other aspects of the research life cycle, such as retrieval and reuse.
  • User interface refinement: While the user interface is being developed iteratively, further refinement is needed to fully meet researcher needs, particularly in areas like dynamic classification and interactive analysis.
  • Capturing adequate granularity and flexibility: The challenge of capturing the right level of detail and adaptability in representing research data remains an ongoing area of development.

Summary updated

BibTeX
@inproceedings{schindler2011how,
  author = {Christoph Schindler and Cornelia Veja and Marc Rittberger and Denny Vrandecic},
  title = {{How to teach digital library data to swim into research}},
  booktitle = {I-semantics},
  year = {2011},
  doi = {10.1145/2063518.2063537},
}
Figure