III: Specification and Analysis of Multi-Actor Data-Driven Business Processes
Businesses and other organizations increasingly rely on electronic workflows
underlying business processes. These workflows are often centered around
a database and involve multiple interacting actors carrying out different roles.
They are typically very complex and prone to costly bugs, which leads
to the need for computer-aided design and analysis tools. These are critical
for increasing confidence in the robustness and correctness of complex
The general objective of this project is to develop new techniques and tools for specifying,
analyzing, managing, and optimizing complex data-driven business processes,
with an explicit focus on the multi-actor setting.
This includes data sharing and updating, designing actor interface specifications to
the global workflow tailored to different roles, and specification of contractual obligations
among interacting actors. It will also consider runtime analysis tasks, including the
distributed monitoring of critical global properties, the use of local actor observations
to infer information about the global workflow, and validating autonomous offline
partial workflow executions by actors. The techniques and tools resulting
from this project will increase confidence in the robustness and correctness
of such systems, and enhance their functionality and efficiency. This will
benefit a wide variety of applications such as supply-chain management,
e-commerce, digital government and electronic health-care management.
The project will contribute to the development of human resources by training graduate
and undergraduate students in cutting-edge specification and static analysis
techniques which are crucial to business process management.
Achieving the objectives of the project requires fundamental advances in our understanding
of data-driven workflows involving multiple interacting actors. In the broader context of
software verification, the goals are particularly challenging because they involve the analysis
of infinite-state systems, whereas classical software verification typically deals with
finite-state abstractions. The project will extend the techniques recently developed for
data-driven workflows to the multi-actor scenario, and provide insight into the trade-off
between expressiveness of the specification language and tractability of analysis tasks.
The technical problems raised are intellectually challenging, and will bring into play
techniques from logic, automata theory, computational complexity, algorithms,
and computer-aided verification.
Alin Deutsch, Yuliang Li, Victor Vianu
PODS 2016, to appear
Serge Abiteboul, Pierre Bourhis, Victor Vianu
Abiteboul, Pierre Bourhis,
Theory of Computing Systems, 2015
Richard Hull, Victor Vianu
SIGMOD Record 43(3), 2014