ProvKOS: A data model for updating knowledge organization systems

Catalogers and researchers often depend on established systems like the Dewey Decimal Classification (DDC) to organize and find information. While these systems have a long history, they are updated regularly to reflect new knowledge and societal changes. However, the process behind these updates—what was changed, why, and based on what evidence—is not visible to the users.  

ProvKOS is a new data model representing an innovative step forward for cataloging and research, offering a way to document and share the story behind every classification update. This transparency enhances trust, usability, and adaptability in knowledge organization systems.

What this research did 

The ProvKOS project explored the context needed to make decision-making processes for updating knowledge organization systems (KOS) more transparent to the user. The information identified as necessary for end users to have access to in a KOS included:

• The historical origins of specific classifications
• When new concepts begin to appear in a system
• The reasons why new concepts are added or existing ones are changed
  

The research then developed a model to document and communicate these updates using structured data so that users can easily access and process the information for further analysis. 

The solution: ProvKOS 

The project introduced a new data model called ProvKOS (Provenance + Knowledge Organization System). This model uses the concept of provenance—commonly used to document the history and origin of objects or data collections—to make the editorial changes in knowledge organization systems, like the DDC, more discoverable and understandable by end users.

ProvKOS captures and organizes information about:

• What caused a change (e.g., societal shifts, new knowledge)
• What evidence supports the change (e.g., references, expert input)
• How a concept in the system has evolved over time (e.g., changes in classification)

By applying provenance to knowledge organization systems, ProvKOS helps users:

• Build trust in the system by making changes more transparent
• Better understand the context and reasoning behind classifications
• Use the system more effectively for research and cataloging

Why it matters 

This innovation has the potential to transform cataloging by creating systems that are more transparent and also better suited to modern technological and societal needs. ProvKOS aligns with FAIR principles (Findable, Accessible, Interoperable, and Reusable), making knowledge organization systems easier to understand and work with. It improves machine readability and supports better querying and data analysis, particularly in linked data environments. 

Collaboration and methodology 

This research was conducted in collaboration with Jessica Yi-Yun Cheng (Rutgers University). The team reviewed open-source models like Prov, ProvONE, SKOS, and SKOS-XL and evaluated DDC data sources such as EPC exhibits, WebDewey, and MARC classification. Using standard ontology development methods (for example, user personas, competency questions, and validation), the team developed and refined the ProvKOS model.

Outputs

Additional Outputs

ProvKOS Ontology (draft)
ProvKOS aims at aiding the management of these more traditional KOSs by providing provenance vocabularies to reflect changes that might otherwise look arbitrary. In total, ProvKOS consists of eight homegrown classes and subclasses and five homegrown object properties, along with five classes and six properties from SKOS and PROV. The use of ProvKOS is exemplified through a case scenario in the Dewey Decimal Classification
https://w3id.org/def/ProvKOS

Project Lead

• Inkyung Choi, OCLC

Project Team

• Jessica Yi-Yun Cheng (Rutgers University, External Partner)
• Alex Kyrios (OCLC, consultant)
• Jeff Mixter (OCLC, consultant)