One of the primary focuses of Semantic Business Process Management is the application of ontology-based semantics for the machine processable representation of business processes and the automation of their management lifecycle. Towards that direction, various ontologies have been proposed, each covering one or more aspects of the knowledge required to describe a business process. Yet, one major limitation of these ontologies is their inability to express knowledge that is vague as they are based on bivalent ontological formalisms. In this context we argue in this paper, through concrete examples and use cases, in favor of using fuzzy ontologies for the effective capture, representation and exploitation of the vagueness that may characterize business processes and we provide initial directions of how this may be practically achieved.
The use of ontology-based semantics for the modeling of business processes is
an emerging research area that aims at creating process descriptions with
explicit and shareable meaning, thus achieving better management of
organizational knowledge and higher level of process automation [
Yet, a dimension of business process knowledge that has so far been inadequately considered within the relevant community is that of vagueness. Vagueness, typically manifested by terms and concepts like Tall, Strong, Expert etc., is a quite common phenomenon in human knowledge and it is related to our inability to precisely determine the extensions of such concepts in certain domains and contexts. That is because vague concepts have typically fuzzy boundaries which do not allow for a sharp distinction between the entities that fall within the extension of these concepts and those which do not. This is not usually a problem in individual human reasoning but it can become one when multiple people need to agree on the exact meaning of such terms and when machines need to reason with them. For example, a system could never use the statement \This process requires many people to execute" in order to determine the number of people actually needed for the process.
In this paper we argue over the need for a systematic way of capturing,
representing and ultimately using vague knowledge in business process management
and we propose for that the utilization of techniques and methods from the area
of fuzzy ontologies [
A basic argument for this is that the modeling inconsistencies that may arise
due to the freedom that business analysts have to name and describe process
knowledge can also occur (and in fact even more severely) when the analysts
need to describe vague concepts and terms. As argued in [
The focus of this paper is on the illustration of the need and applicability of fuzzy ontologies in vague business process management so that the foundations for future research on this area may be set. Therefore in section 2 we draw and analyze examples from state of the art business process ontologies and related application scenarios in order to highlight the forms in which vagueness may be present in semantic business process information and the need for a formal treatment of it. In section 3 we describe the notion of fuzzy ontologies and we show how these may be developed and used for business process modeling by means of relevant state of the art methods and techniques. Finally, in the last section conclusions and directions for future work are provided. 2 2.1
Vagueness as a semantic phenomenon is typically manifested through predicates
that admit borderline cases [
It should be noticed that vagueness is different from inexactness or uncertainty. For example, stating that someone is between 170 and 180 cm is an inexact statement but it is not vague as its limits of application are precise. Similarly, the truth of an uncertain statement, such as \Today it might rain", cannot be determined due to lack of adequate information about it and not because the phenomenon of rain lacks sharp boundaries.
In an ontology the elements that can be vague are typically concepts,
relations, attributes and datatypes [
The term “process knowledge” refers to the information describing the control flow of a process as well as its content, namely all artifacts that its definition may refer to. These artifacts are typically derived from and express the business environment and the organizational context of the process. Vague pieces of information and knowledge may appear in all three dimensions of process knowledge, namely structure, domain and organizational context.
To illustrate this point, we have considered and analyzed, with the help of our company’s consultants, two different cases of business process knowledge. The first case involved a set of generic business process related ontologies, developed in project SUPER 1, which may serve as reusable knowledge schemas in practical semantic business process modeling scenarios. The analysis of these ontologies, which included among others the Business Process Modeling Ontology (BPMO), the Business Goals Ontology (BGO), the Business Roles Ontology (BROnt) and the Business Motivation Ontology (BGO), involved the identification within them of elements that can be interpreted as vague, according to the definitions of the previous paragraph. Our criterion for classifying an element as vague or not was merely the potential existence of borderline cases, not the number of them. That meant that even if an element could potentially have only one borderline case, it was considered vague.
The outcome of this analysis is summarized in table 1 where a sample of the elements we managed to identify as vague, along with a brief explanation of their vagueness, is presented. As one can easily see, the identified as vague elements are quite central to their respective ontologies (e.g. the hasBusinessGoal relation) and as such they are expected to be found in many relevant application scenarios. Furthermore, the use of vague terms like “desired” in the definition of elements (e.g “Desired Result”) indicates that in practice there could be an almost infinite number of vague ontological elements in these ontologies that would be the result of the combination of such terms with non-fuzzy elements (e.g. “Loyal Customer”, “Expert Analyst” etc.).
Vagueness Combinatory vagueness due to the lack of sharp discrimination between those conditions that are necessary for someone to be considered as having a managerial role Degree-vagueness along the dimensions of the number of competitor’s business areas and target markets that constitute someone as a competitor Combinatory vagueness due to the lack of sharp discrimination between those conditions that are necessary for something to belong to a given domain Combinatory vagueness due to the lack of sharp discrimination between those conditions that are necessary for a goal to be strategic Combinatory vagueness when criteria for desirability have not been set or are vague, degreevagueness when these criteria are arithmetic
The second case of business process knowledge we considered involved a
specific application scenario, derived from [
Our analysis of this process involved identifying which aspects of it (structure, domain knowledge etc) had vague characteristics. Our findings can be summarized as follows: First, some of the process’s various decision conditions according to which a specific action is decided are vague. For example, in order to take the decision about pursuing the call, two criteria that need to be satisfied are i) the budget of the project to be high and ii) the company’s experience to be adequate. In both cases there could be borderline cases as it is indeterminate what is the exact threshold over which the budget is considered high (degree vagueness) or how many years and how many projects are required exactly for the company to be considered experienced in a given area (degree vagueness in two dimensions). Second, many of the underlying organizational and domain pieces of knowledge that are needed for performing various steps of the overall process are also vague. For example, the assessment of the potential competition for the call requires knowledge about the company’s competitors. Yet, the existence of other companies that are borderline competitors is possible, mainly due to the lack of clear criteria about what constitutes a competitor and what not (combinatory vagueness). A similar argument can be made for the knowledge about the company’s areas of expertise.
This second case illustrates, apart from the existence of vagueness in a common business process, the potential problems that may be caused during the latter’s execution when this vagueness is not formally considered. Different people who will perform the same process will most likely produce different results, exactly because they will interpret various pieces of knowledge in a different manner (e.g. what budget is considered “high” or which companies are competitors). And it should be noted that this is not merely a problem of inadequate measurement or lack of concrete business rules but an inherent problem caused by the vagueness of human knowledge. For example, even if there is a business rule suggesting that competitors are those who have clients in the same industries and services in the same areas the question remains: what is the minimum number of similar clients or services that a given company needs to have in order to be considered a competitor? In the next section we describe how fuzzy ontologies may be practically applied for dealing with questions like that. 3
A fuzzy ontology utilizes notions from Fuzzy Set Theory in order to formally represent the vague ontological elements described in paragraph 2.1. The basic elements it provides include i) Fuzzy Concepts, namely concepts to whose instances may belong to them to certain degrees (e.g. Goal X is an instance of StrategicGoal at a degree of 0.8 ), ii) Fuzzy Relations/Attributes, namely relations and attributes that link concept instances to other instances or literal values to certain degrees (e.g. John is expert at Knowledge Management to a degree of 0.5 ) and iii) Fuzzy Datatypes, namely sets of vague terms which may be used within the ontology as attribute values (e.g. the attribute experience mentioned above). In a fuzzy datatype each term is mapped to a fuzzy set that assigns to each of the datatype’s potential exact values a fuzzy degree indicating the extent to which the exact value and the vague term express the same thing (e.g. a consultant with 5 years of experiences is considered junior to a degree of 0.4 )
As with classical ontologies, using and applying fuzzy ontologies in
practical scenarios requires corresponding methods and tools for developing them,
formally representing them and performing reasoning and querying over them.
Two recent developments that may cover a great part of these requirements is
the IKARUS-Onto [
To assess the feasibility and potential value of our approach we applied it in developing a fuzzy ontology for the aforementioned tender call evaluation process. Due to space limitations we cannot describe this ontology in detail, nevertheless it suffices to say that pieces of knowledge with degree-vagueness were typically modelled as fuzzy datatypes (e.g. budget and experience) while those with combinatory vagueness (e.g. competitors) were modelled as fuzzy relations or concepts. Figure 3 depicts two sample definitions in Fuzzy OWL 2.
In any case, the developed fuzzy ontology formed the basis of a simple decision support system which queried the ontology in order to evaluate the tender call evaluation criteria and provide a suggestion to its users on whether they should pursue a given call. We had this system used by some of our consultants and we asked from them to provide some informal feedback. The positive remarks we received regarded the explicitness of the fuzzy elements’s meaning (a result of using IKARUS-Onto) as well as the automation achieved in the retrieval of vague knowledge that made easier the evaluation of the fuzzy decision criteria. On the other hand, the task of defining fuzzy degrees and membership functions was deemed as quite difficult and time-consuming, indicating thus the need for more automated methods for vague knowledge acquisition. 4
In this paper we utilized concrete examples and use cases from the literature in order to highlight the omnipresence of vagueness in business process knowledge and to argue in favour of using fuzzy ontologies for dealing with it in a semantically rich manner. Furthermore, we provided a small but representative example of how some of the latest fuzzy ontology engineering methods and tools may be practically used for dealing with vagueness in business process modeling. In the future we intend to further substantiate and support the use of fuzzy ontologies in semantic business process modeling by providing dedicated methods and tools for the capture, management and exploitation of vague process knowledge and by applying and evaluating these methods in real application scenarios.
The research leading to this work has received partial funding from the ECs FP7 programme: (FP7/2011-14) under grant agreement no. 287615 (PARLANCE).