Laura Sánchez-González

Prediction of business process model quality based on structural metrics

The quality of business process models is an increasing concern as enterprise-wide modelling initiatives have to rely heavily on non-expert modellers. Quality in this context can be directly related to the actual usage of these process models, in particular to their understandability and modifiability. Since these attributes of a model can only be assessed a posteriori, it is of central importance for quality management to identify significant predictors for them. A variety of structural metrics have recently been proposed, which are tailored to approximate these usage characteristics. In this paper, we address a gap in terms of validation for metrics regarding understandability and modifiability. Our results demonstrate the predictive power of these metrics. These findings have strong implications for the design of modelling guidelines.

Quality indicators for business process models from a gateway complexity perspective

Context: Quality assurance of business process models has been recognized as an important factor for modeling success at an enterprise level. Since quality of models might be subject to different interpretations, it should be addressed in the most objective way, by the application of measures. That said, however, assessment of measurement results is not a straightforward task: it requires the identification of relevant threshold values, which are able to distinguish different levels of process model quality. Objective: Since there is no consensual technique for obtaining these values, this paper proposes the definition of thresholds for gateway complexity measures based on the application of statistical techniques on empirical data. Method: To this end, we conducted a controlled experiment that evaluates quality characteristics of understandability and modifiability of process models in two different runs. The thresholds obtained were validated in a replication of the experiment. Results: The thresholds for gateway complexity measures are instrumental as guidelines for novice modelers. A tool for supporting business process model measurement and improvement is described, based on the automatic application of measurement, and assessment as well as derivation of advice about how to improve the quality of the model. Conclusion: It is concluded that thresholds classified business process models in the specific level of understandability and modifiability, so these thresholds were good and useful for decision-making.

Towards thresholds of control flow complexity measures for BPMN models

Business process models are considered to be a good mechanism for communication among stakeholders and are a key instrument in the analysis and design of information systems. It is therefore important to design business process models with a high level of quality, which can be discovered through measurement application. Several measurement initiatives exist in the literature, but these measures are only useful in real world decision making if we also have criteria with which to establish the goodness of models. We consider that measures with thresholds and decision criteria form indicators. Indicators allow us to make decisions by using the values of the measures which models should not exceed to ascertain whether the model is good in practice. In this paper we present the initial empirical results from which thresholds for the Control-Flow Complexity measure applied in BPMN models have been obtained according to the Bender method. Our findings reveal that there are different levels of understandability depending on the number of decision nodes: a very easily understandable model would have no more than 6 xor nodes, 1 or nodes and 1 and nodes, versus the 46 xor nodes, 14 or nodes and 7 and nodes which would constitute a model with a very difficult level of understandability.

Quality assessment of business process models based on thresholds

Process improvement is recognized as the main benefit of process modelling initiatives. Quality considerations are important when conducting a process modelling project. While the early stage of business process design might not be the most expensive ones, they tend to have the highest impact on the benefits and costs of the implemented business processes. In this context, quality assurance of the models has become a significant objective. In particular, understandability and modifiability are quality attributes of special interest in order to facilitate the evolution of business models in a highly dynamic environment. These attributes can only be assessed a posteriori, so it is of central importance for quality management to identify significant predictors for them. A variety of structural metrics have recently been proposed, which are tailored to approximate these usage characteristics. The aim of this paper is to verify how understandable and modifiable BPMN models relate to these metrics by means of correlation and regression analyses. Based on the results we determine threshold values to distinguish different levels of process model quality. As such threshold values are missing in prior research, we expect to see strong implications of our approach on the design of modelling guidelines.

TOWARD A QUALITY FRAMEWORK FOR BUSINESS PROCESS MODELS

Business process modeling is recognized as a key part of the business process lifecycle. It is during this stage that a conceptual model is produced by collecting business process requirements and ...

A study of the effectiveness of two threshold definition techniques

Background: Measurement is a technique that is widely-used to quantify quality of process models. Evaluation of measurement results implies comparison against limit values, called thresholds. Determining thresholds is no trivial task and it requires the application of complex techniques. There are several techniques that have been published to date, proposing different approaches for threshold extraction. Two of the most prominent techniques are ROC curves and the Bender method. Although they come from different fields, both use logistic regression analysis as a discriminator function. Aim: For this reason, the main hypothesis is that thresholds obtained by both of those techniques are equally efficient in classifying the measurement results. Method: To check the hypothesis, we obtained thresholds for a group of empirically-validated measures for business process models, by applying both techniques. Then we checked the accuracy of the results. Results: The results indicate that the hypothesis should be rejected. Conclusions: ROC curves obtained more accurate thresholds for measurement evaluation.

Improving Quality of Business Process Models

Business process improvement is a key aspect for organizational improvement. We focus the business process improvement in the first stage of process lifecycle, design stage, because it is a means to avoid the propagation of errors to later stages, in which their detection and correction may be more difficult. Since business process improvement is centered in business process models, a proposal of certain steps based on measurement activities on conceptual models (measurement, evaluation and redesign) is described. The application of these steps in business process models produces an increase of the quality of them. Quality is defined as the level of understandability and modifiability, subcharacteristics of the usability and maintainability in ISO 9126. The steps for model improvement have been applied to a real hospital business process model. The model was modified by following expert opinions and modeling guidelines, thus leading to the attainment of a higher-quality model. Our findings clearly support the practical utility of measurement activities for business process model improvement.

A case study about the improvement of business process models driven by indicators

Organizations are increasingly concerned about business process model improvement in their efforts to guarantee improved operational efficiency. Quality assurance of business process models should be addressed in the most objective manner, e.g., through the application of measures, but the assessment of measurement results is not a straightforward task and it requires the identification of relevant indicators and threshold values, which are able to distinguish different levels of process model quality. Furthermore, indicators must support the improvements of the models by using suitable guidelines. In this paper, we present a case study to evaluate the BPMIMA framework for BP model improvement. This framework is composed of empirically validated measures related to quality characteristics of the models, a set of indicators with validated thresholds associated with modeling guidelines and a prototype supporting tool. The obtained data suggest that the redesign by applying guidelines driven by the indicator results was successful, as the understandability and modifiability of the models were improved. In addition, the changes in the models according to guidelines were perceived as acceptable by the practitioners who participated in the case study.