Towards Automating Inter-organizational Workflow Semantic Resolution

Interoperability in sharing work through services between organizations requires understanding the perspective of each partner. Available resources and requirements are widely distributed with heterogeneous descriptions and enactments. No centralized registration and discovery process exists, nor automated means to map the variety of formal organizations within various overlapping domains, some defined and many not. While drawing from a common language, the mission and culture of each organization shapes their particular collection and definition of symbols, vocabulary and signals. Mining execution logs of historical inter-organizational workflows and tapping into explicitly specified domain and organizational knowledge ontologies provides a corpus of information useful to identify emergent patterns. This paper proposes a novel mediator approach using feature alignment demonstrated in previous partnerships, identified through the employment of topic modeling and word sense disambiguation methods, to infer semantic matches between parties without a priori relationships. This model demonstrates an inter-organizational workflow middleware proof of concept coupling workflow management systems interoperating between a set of organizations to automatically resolve meaning, providing the right services for requirements.