Business Impact Analysis aims at improving a companies business processes by supplying information to an analyst. The analyst's task is to write queries which bridge that gap between a BPEL process and a data warehouse, as information from both sources are needed.
To provide tool support to the analyst, the BIA Annotation Editor was developed during previous work. The initial version of the editor contained the basic functionality towards a semi-automatic semantic linking method. It allowed the user to annotate the process variables and the data warehouse tables and columns with concepts taken from an ontology.
This thesis directly ties with this editor in that it provides even more tool support by implementing the semi-automatic semantic linking method using the annotations inserted using the BIA Annotation Editor. The IntraOntologyMatching (IOM) was inserted into the editor. The user interface was extended. The initial functionality was preserved on its own perspective, whereas the IOM results are displayed on a new perspective. In addition to the IOM, which is inherently semi-automatic in that it needs the users intervention to produce matches, a schema matching algorithm was implemented and integrated into the BIA Annotation Editor. The schema matching algorithm can be executed without further ado and it directly produces matches. It is based on Cupid and extends it for matching process variables to operational data models. It is used for the OSM and the PSMA.
The OSM is used to match the ontology onto the data warehouse, which is done without any semantic connotation and which is strictly based on the schema matching approach. It is useful, if only the process is semantically annotated which is a very likely situation to be encountered in real world scenarios. Matches are created from the annotations onto the columns and tables of the data warehouse via Cupid element level schema matching. All of a concept´s matches are then copied onto the process elements that were annotated using that concept. Therefore OSM ultimately finds matches between the process and the data warehouse.
PSMA performs element level and structural level matching. It is novel in that it understands the structure of a BPEL process. Given the files that define the process it can reconstruct the complete process structure, decide which process elements to match and which elements to use as sources of additional information only. The advantage the BIA Annotation Editor provides over other schema matching tools is that it is specialized on BPEL processes. It matches only the process variables to the data warehouse, which results in the meaningful subset of the matches a generic tool would produce given the same input. PSMA exploits the variables context and connects variables to each other based on a data flow analysis of the process. Match hints given by the context and by connected variables are then exploited to enrich the overall result.
After a chapter describing the functionality and its implementation, the IOM, OSM and the PSMA have been evaluated in this thesis. It shows that user intervention is still necessary as the matching results are not perfect. The user still has to decide which matches are correct and which matches he wants to ignore. On the other hand, the results are promising. The analyst using the BIA Annotation Editor receives helpful suggestions via the editors graphical user interface concerning elements that match between the BPEL process and the data warehouse, which enable him to navigate the schemas he is confronted with more easily, quickly and more safely. He can perform his ultimate task of designing the queries needed for the Business Impact Analysis as he receives the desperately needed tool support via the BIA Annotation Editor.
|