Automatic Generation of Optimized Business Process Models From Constraint-Based Specifications

Abstract

Business process (BP) models are usually defined manually by business analysts through imperative languages considering activity properties, constraints imposed on the relations between the activities as well as different performance objectives. Furthermore, allocating resources is an additional challenge since scheduling may significantly impact BP performance. Therefore, the manual specification of BP models can be very complex and time-consuming, potentially leading to non-optimized models or even errors. To overcome these problems, this work proposes the automatic generation of imperative optimized BP models from declarative specifications. The static part of these declarative specifications (i.e. control-flow and resource constraints) is expected to be useful on a long-term basis. This static part is complemented with information that is less stable and which is potentially unknown until starting the BP execution, i.e. estimates related to (1) number of process instances which are being executed within a particular timeframe, (2) activity durations, and (3) resource availabilities. Unlike conventional proposals, an imperative BP model optimizing a set of instances is created and deployed on a short-term basis. To provide for run-time flexibility the proposed approach additionally allows decisions to be deferred to run-time by using complex late-planning activities, and the imperative BP model to be dynamically adapted during run-time using replanning. To validate the proposed approach, different performance measures for a set of test models of varying complexity are analyzed. The results indicate that, despite the NP-hard complexity of the problems, a satisfactory number of suitable solutions can be produced.

Generating optimized configurable business process models in scenarios subject to uncertainty

“…Each activity act is a tuple <actid; Pred; dur; es; le; res> where: actid is an unique identifier in the enactment plan, Pred is the list of its precedence activities (i.e., those activities which must be executed before act), dur is the estimated duration of act, es is the earliest start time (i.e., the soonest that the activity act can start), le is the latest end time (i.e., the latest that the activity act can finish), and res is the resource which performs the activity act in the plan. 7 Such definition is provided to formalize the concepts which already exist in the literature related to that term.…”

Automatic Generation of Questionnaires for Supporting Users during the Execution of Declarative Business Process Models

“…For supporting users working on declarative BP models, we proposed in previous works [2,1] an approach for generating an optimized execution plan (i.e., an optimized variant) from a given declarative BP model at design time (cf. Fig.…”

“…black dot in Fig. 1 (a)) [1,10], and hence, diminishes the advantages of using declarative models. In particular, if BPs are subject to uncertainty and conditions may change during the BP execution, it might turn out that the selected variant is not applicable and replanning might be required.…”

“…To support users working on declarative BP models while a high flexibility is maintained, we propose unlike [2,1] to not select a single variant, but to keep the best variants before the BP enactment. For managing these variants during runtime, they can be automatically merged into a configurable BP (CBP) model (i.e., a modeling artifact that captures families of BP models in an integrated manner [16,15]) by using the techniques presented in [9].…”

“…[11]). To facilitate the human work involved in such design, to avoid failures, and to allow for a better optimization during the execution phase [7], declarative BP models are increasingly used allowing their users to specify what has to be done instead of how [1,10]. However, due to their flexible nature, there are frequently several variants related to a given declarative model, each one presenting specific values for different objective functions (e.g., overall completion time or profit).…”