Pengembangan Aplikasi Manajemen Proyek Perangkat Lunak Menggunakan Framework Serum (SPM-Serum)

Studi Eksplorasi terhadap Praktik Requirements Engineering dalam Proses Pengembangan Perangkat Lunak di Indonesia

P-Gamelan* : Process Ecosystem Views to Managing Changes in Business Process Repositories

A business process is a collection of activities designed to provide an output for a particular customer or market. It implies a strong emphasis on how the work is done within an organization, in contrast to a product’s focus on what. A process is thus a specific ordering of work activities across time and place, with a beginning, an end, and clearly defined inputs and outputs: a structure for action. Most medium to large organizations support large collections of process designs modeled through various business process modeling languages such as Business Process Model and Notation (BPMN). They are usually stored in business process repositories. These are typically characterized by the following features: (i) the number, scale and complexity of the processes are large (i.e. hundreds to thousands of models); (ii) the processes are often interdependent; (iii) changes to any one process are likely to impact several other processes and (iv) there are multiple alternative ways in which changes might be implemented.

Dealing with such complex process repositories is not a trivial task. In this research, we argue that a formal process ecosystem view can provide a particularly useful solution to the problem. In our conception of process ecosystem, process designs play a role analogous to that of biological entities in a biological ecosystem (i.e. they can be created, modified, eventually discarded). As biological ecosystems always try to get their equilibria, this framework maintain inter-process dependency constraints to get process ecosystems equilibria. Hence, P-Gamelan framework relies on such metaphor to represent a choreography of processes within a repository, analogous to a set of instruments choreography in a gamelan. This consolidating view of the complete landscape of business processes would yield some advantages for the organizations, e.g. avoiding inconsistencies between process models, achieving global optimization among the related process models.

Obviously, process analysts need a suitable supporting tool to perform such complex business process management. Some useful BPM tools have been developed carefully and used by some companies for managing their business process models, e.g. ARIS¹, APROMORE², ADONIS³. They have partially or entirely the following features including creation, modification and deletion, filtering/searching and evaluation of the process models. However, there are still open issues to be addressed: (i) process repository management requires the establishment and maintenance of a repertoire of relationships between process designs, such as generalization-specialization, part-whole, inter-operation, abstraction-refinement and resource-sharing; (ii) such repository may also contain designs that are partial or incomplete that need to be detected and rationalized; (iii) process improvements may lead into a contradictory situation among related process models which requires a particular resolution and (iv) changes on a single process model likely need to be propagated to the other related models in order to maintain their relationships.

This research will attempt to deal with such issues. We build the P-Gamelan framework based upon semantic effect-annotated process models represented in BPMN, i.e. process models in which every activity has been annotated with its corresponding effects. An effect annotation relates to a particular result or outcome to an activity in a process model. Hence, we are interested in the following research questions: (i) how can we formalize inter-process relationships in semantic effect-annotated business process models?; (ii) how can we maintain the normative inter-process relationship between a pair of process models?; (iii) how can we maintain the equilibrium of a process ecosystem due to process changes? and (iv) how can we facilitate the process analysts in dealing with process optimization?. We will provide the P-Gamelan with the following key capabilities, i.e. process repository management, process rationalization, process optimization and process change management including process redesign. By developing this framework, we do not intent to replace the role of process analysts, but we will assist them with an analyst-mediated process management tool.

* P-Gamelan stands for 'Process Gamelan'. Gamelan is a traditional musical ensemble from Indonesia, featuring a variety of instruments. It is composed of a number of instruments as a distinct object, arranged and tuned to stay together.

1. http://www.ids-scheer.com/
2. http://apromore.org/
3. http://www.boc-group.com/products/adonis/

Semantically Effect Annotated Business Process Model

A business process is a collection of activities designed to provide an output for a particular customer or market. It implies a strong emphasis on how the work is done within an organization, in contrast to a product’s focus on what. A process is thus a specific ordering of work activities across time and place, with a beginning, an end, and clearly defined inputs and outputs: a structure for action. This usually is represented in diagram using particular business process modeling techniques, e.g. Business Process Model and Notation (BPMN). The current tools supporting such modeling techniques focus on the process ordering and coordination semantics of business process. We will not be able to determine, at design time, the effects of process execution up to a certain point in the model. Hence, we extend the current process modeling techniques with the semantic effect annotation on business processes.

An effect annotation relates to a particular result/outcome to an activity in a process. An activity represents the work performed within a process. Activities are either atomic (called a task) or compound (called a sub-process). In an annotated BPMN process model, as our approach relies on, we annotate each activity with its (immediate) effects. We define the immediate effects as the immediate results/outcomes of executing an activity in a process. This annotation allows us to determine, at design time, the effects of process execution up to a certain point in the model. These effects are necessarily non-deterministic, since a process might have taken one of many possible alternative paths through a model to get to that point, called as effect scenarios. We define a procedure for pair-wise effect accumulation, which, given an ordered pair of activities with their corresponding effect annotations, determines the cumulative effects after both activities have been executed in a contiguous sequence. In a process model, there would be many effect scenarios which are introduced by AND-joins or XOR-joins or OR-joins. We accumulate effects through a left-to-right pass of a participant lane, applying the pair-wise effect accumulation procedure on contiguous pairs of activities. We use Answer Set Programming (ASP) theorem prover for reasoning the effects being accumulated. Note that we do not consider the possibility of a pair of effect scenarios in AND-joins being inconsistent, since this would only happen in the case of obviously erroneously constructed models.