7.4 Semantic Execution Environment
7.4.3 Execution Principle
In order to execute the generated process, a special execution environment is needed, in which SISi is only one part. Additionally, the landscape includes an orchestration engine for executing the generated BPEL process as well as a web service engine hosting the involved web services. Furthermore, also a semantic reasoner is needed for discovering the web services based on the given semantic descriptions. The overall system landscape and the underlying execution principle can be found in figure 7.9. The execution of the BPEL process works as follows:
1. The BPEL process is executed on a standard orchestration engine. Whenever a semantically annotated function is found, the request is forwarded to SISi.
2. SISi receives the semantic discovery request. Besides the semantic description, SISi also takes the needed input parameters. SISi passes the semantic description to the semantic reasoner.
3. The reasoner uses semantic discovery algorithms to find matching semantic web service descriptions. The best fitting web service description is selected and passed back to SISi.
4. SISi uses the semantic web service description and finds through the included grounding information the underlying concrete web service implementation. Then, this web service is invoked with the input parameters received in the second step. 5. The web service is executed and the output data returned to SISi.
6. SISi forwards the output back to the BPEL process.
In the semantic business process management application, the system architecture is implemented with the following software components:
• The semantically annotated BPEL process is executed on the Oracle BPEL Process Server4. The BPEL process definition includes invocations of SISi each time a web service must be discovered during runtime.
• SISi itself is hosted on an instance of Apache Tomcat5.
• WSMX6is used as semantic discovery environment. WSMX includes the necessary reasoning components. WSMX also supports invoking the discovered web services. Therefore, WSMX is doing the actual invocation of the web services and passes the output message back to SISi. Step 4 and 5 are handled by WSMX as shown in figure7.9.
• The web services are hosted on another instance of Apache Tomcat.
4http://www.oracle.com/lang/de/technologies/soa/soa-suite.html
5http://tomcat.apache.org/
8
Case Study:
Semantic Business
Process
Management
This thesis uses case studies to evaluate the artefacts created. This chapter presents the second case study, which mainly focuses on evaluating the ARIS extension (see chapter3) and the semantic business process management application (see chapter7). This chapter is structured as follows: First, the motivation for this case study is presented in section8.1. Afterwards, the research aim is presented in section8.2. The scenario used in the case study is presented in section8.3. The case study done is described in section8.4and the results are presented and discussed at the end of this chapter in section8.5.
8.1 Introduction
Semantic business process management as summarised in section2.6 promises many advantages by enabling machines to process the content of enterprise models. Re- searchers aim at making semantic business process management a complete modelling method by working on specific modelling procedures, modelling languages, and applica- tions. However, as the evaluation of semantic business process management literature in subsection2.6.6shows, today no empirical studies are available to validate the usefulness of it. In addition, semantic business process management is only expected to be suc- cessful if existing models (e. g. business process models) can be reused. So far, work on ontologies built on top of existing modelling languages is provided, but a more integrated approach is not available.
The present case study tackles both problems: providing empirical evaluation of seman- tic business process management and integrating it with an existing modelling method for business process management. In context of the ARIS extension developed, the case study evaluates if the ARIS extension is flexible enough to be combined with a more for- malised approach. The case study presented is using the semantic business process management application described in chapter 7. Also the EPC to BPEL transformation application is reused and extended as described in subsection7.3.5.
SUPER1. As such a research project is a collaborative effort, work done by the use-case partner Telekomunikacja Polska [ESF+08, pp. 19] was reused. The business process of the use-case partner TP was converted into an EPC model and adapted to the needs of the case study. In addition, WSMO goals as described in [FRS07] were reused without any changes, too. Reusing this existing work ensured that the example process used in the case study is a realistic one, which increases the validity of the case study. The collaboration between Telekomunikacja Polska and the author is documented in joined publications [SSEKR08,SSEKR09].
8.2 Research Aim
This case study aims at evaluating semantic business process management in an indus- trial setting. Therefore, the following research question is defined: “How does business process management benefit from introducing semantic technologies and how is the adop- tion in industry of semantic business process management hindered?”
The research interests are of explorative nature. According to Yin [Yin03], controlled experiments as well as case studies are research methods able to answer such “how” and “why” questions. Kitchenham et al. [KPP95] add that experiments are usually applied for “research-in-the-small” and case studies for “research-in-the-typical”. The research aim clearly focuses on research-in-the-typical, because it intends to investigate the usage of semantic business process management in a real-world setting. Kitchenham et al. also state that case study research is often used to evaluate new technologies. This also ap- plies for the research aim defined. To ensure that the research is conducted in an realistic environment, the existing ARIS method is used and extended to support semantic busi- ness process management as discussed in chapter7. In addition, the case study is based on a real-world use case, which is described in section8.3.
The case study research follows the methodology defined by Yin [Yin03]. Yin’s method- ology is augmented with ideas taken from Kitchenham et al. [KPP95], because they de- scribe specific practices for case study research in software engineering, which are appli- cable here as well.