As business processes are increasingly complex, agile, and specialized, provenance information can improve interpretability and contextualization of process data. While individual process steps frequently employ digital computer files, their relationships within the overall process are rarely captured. To address this issue, we extend the factFUSE system for managing versioned Web resources in the file system to capture provenance relationships. By introducing an extensible commit system, we enable recording the relations between digital files and resources in process steps (activities), which are then captured as RDF metadata using the W3C PROV standard. Our evaluation shows that users without prior experience in provenance management successfully employ the system to capture semantic process provenance, attesting to excellent usability and promising utility. factFUSE is available for practical use under open source GNU AGPLv3 license.
Contributions. Addressing these goals, we propose a platform-independent concept for provenance capturing in the file system and provide a corresponding open-source implementation for both Linux and macOS. We further provide an evaluation of the system w.r.t. the design goals defined above. Sec. 2 introduces the main concept and presents the implementation of an extensible commit system for factFUSE. Sec. 3 discusses its quantitative and qualitative evaluation. We conclude our work in Sec. 4. 2
To capture provenance information on computer files and to enable context embedding
in traditional data management environments, a bridge between the classic hierarchical
file system and semantic data management is fundamental. We extend upon the recently
proposed factFUSE [
Concept. Extending upon this foundation provided by the factFUSE system, we
embed provenance management into the traditional workflow of managing files in the
file system by introducing (i) a Commit system inspired by the distributed version control
system GIT [
Fig. 1. The factFUSE concept, extended to capture computer file provenance through a commit system and corresponding metadata interfaces in the file system. Mountpoint
EDITED RDF1.ttl
C2
IMG1.jpg DELETED RDF2.ttl
Commit
EditA used: IMG1 "Comment" contains RDF1 revisionOf RDF1
generatedBy C1 contains
C2 associatedWith
used lmueller
EditA Comment contains IMG1 collected provenance information is then persisted to the resource’s metadata record to track the changes and revisions that resources go through and capture process knowledge.
Realization. factFUSE is a NodeJS application, providing a custom user-space
file system based on the FactStack [
A commit (modeled as a PROV activity) carries information on the time and content of changed resources (PROV entities), a title, its author (PROV agent), a message, and possibly additional resources that were used (but not modified) in the process. Fig. 2 shows an example in which two resources in the local file system are modified. The CommitGUI illustrated in Fig. 3 is then used to capture provenance and additional metadata of the generating process and to commit the data to the upstream LDP server.
In order to enable the semantic exploration and management of existing resources, a set of user interfaces (see Fig. 4) Fig. 3. Configuring changes & metadata to inhas been designed and implemented, that clude when creating a process step commit. are accessible through the right-click context menu, extending the file system’s functionality. The RevisionView displays a history of all existing revisions of a resource as well as each revisions generating activity. Additionally, UI elements to download or restore revisions are provided. Inspecting a revision’s generating activity allows further inspection along the edges of the PROV graph, by opening the ActivityView. An activity’s ActivityView holds information on its responsible Agent, the time of execution, and a list of resources that have been used during the activity. A full overview of the implemented interfaces in in-use scenarios can be found on the projects repository page4. This approach of providing context-dependent UI extensions to manage and visualize metadata, in alignment with individual ontologies and best practices of specific application domains, allows for the progressive and configurable adoption of semantic data management to the traditional file system.
Along with the proposed provenance management extensions, factFUSE provides a
tool to manage version-controlled Web resources in the file system which automatically
collects provenance information with interfaces to explore and use them. Goal G1 is
met by design since factFUSE allows for the drag-and-drop integration of arbitrary
computer files and its representation of Web resources in the file system that can be used
and edited in existing desktop applications. Goals G2 and G3 are met by employing
the FactStack data management system as detailed in [
In this paper, we presented a solution for the collection and management of provenance information in the file system, by extending the factFUSE system for managing versioned Web resources. Through the implementation of an extensible commit system, we provide a solution for semi-automatic provenance collection combined with manual user prompts for additional metadata that is expressed as RDF, using the W3C PROV standard. This enables the semantic embedment of digital files into processes and subsequently enables a detailed overview of the relations within a process and between different resources. Our user study yielded excellent usability results and showed a quick adoption of principles by new users. As such, factFUSE provides a first step towards the flexible integration of traditional file-based data management with semantic metadata using open Web standards and technologies and is available open-source for community discernment. Acknowledgments Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy – EXC-2023 Internet of Production – 390621612.