I was hoping to gain a deeper understanding of the difficulties of making workflow management systems (WfMS) more context aware. For instance, simply modeling potentially changing context information should not be trivial (think of identity, roles, neighbors, asynchronous events, places, and time), and then the WfMS has to link this context data to the typical (often static) workflow models used in actually executing a workflow. Regrettably, this paper is an extraordinary disappointment in all respects.

The so-called survey merely consists of seven superficial, incoherent, unsystematic, and short (about four to six small paragraphs) descriptions of WfMS allegedly addressing this domain. These snippets resemble product descriptions rather than grounded research work. The authors do not even introduce a common framework or structure for doing the comparison, but simply list various software modules used, the idiosyncratic modeling languages employed (for example, pvPDL, OWL, PNML, UML, uWDL), and other introductory (almost marketing type) information.

In the introduction, the authors try to discuss the problem of defining “workflow” by citing utterly outdated definitions (mid-1990s!) and do not even provide their personal definition. This line of errors is continued by referring to BPEL4WS, WSFL, and XLANG as prototypical workflow execution languages. Besides the fact that both WSFL and XLANG have been actually used to specify BPEL4WS (they never played any substantial role by themselves), the latter has now become obsolete with the creation of BPMN 2.0 and CMMN 1.0.

Instead of discussing the highly ambiguous term “context-awareness,” the authors simply reiterate a (much-too-short) definition from another paper.

Finally, the two major conclusions of the work are either flatly wrong or the authors have simply not provided information substantiating their bold claims. For instance, they purport that there is “no way [for a WfMS] to […] take into account these changes [i.e., of context information] in the ongoing workflow without disturbing its execution.” This is evidently false for BPMN 2.0, which is not even a dedicated context-aware workflow language.

My recommendation: try a Google search on “context-aware workflow systems” rather than invest time to make sense out of this paper.