Interrelationships of Service Orchestrations.

Despite topic models have been successfully used to reveal hidden orchestration patterns from service logs, the potential uses of their interrelationships have yet to be explored. In particular, the popularity of an orchestration pattern is a leading indicator of other orchestrations in many situations. Indeed, the research in capturing relationships by induced networks has been active in some areas, such as in spatial problems. In this paper, we propose a structure discovery process to reveal relationship networks among service orchestrations. In practice, more robust business logic can be formulated by having a good understanding of these relationships that leads to efficiency gains. Our proposed interrelationship discovery process is performed by a set of optimizations with adaptive regularization. These features make our proposed solution efficient and self-adjusted to the dynamics in service environments. The results from our extensive experiments on service consumption logs confirm the effectiveness of our proposed solution.