WhatweknowandwhatwedonotknowaboutDMNkathrinfigl.com/wp-content/uploads/2018/04/What-we...Business Process Model-BPMN Credit Sales Project Specifications Credit Eligibility Rules Credit

Enterprise Modelling and Information Systems ArchitecturesVol. 13, No. 2 (2018). DOI:10.18417/emisa.13.2What we know and what we do not know about DMN 1Research Note

What we know and what we do not know about DMN

Kathrin Figla, Jan Mendling*,b, Gul Tokdemirc, Jan Vanthienend

a University of Innsbruck, Austriab Wirtschaftsuniversität Wien, Vienna, Austriac Cankaya University, Ankara, Turkeyd KU Leuven, Belgium

Abstract. The recent Decision Model and Notation (DMN) establishes business decisions as first-classcitizens of executable business processes. This research note has two objectives: first, to describe DMN’stechnical and theoretical foundations; second, to identify research directions for investigating DMN’spotential benefits on a technological, individual and organizational level. To this end, we integrateperspectives from management science, cognitive theory and information systems research.

Keywords. DMN • BPMN • Process Modeling

Communicated by A. Koschmider. Received 2017-06-18. Accepted after 2 revisions on 2018-03-14.

1 Introduction

The Decision Model and Notation (DMN) is arecent standard of the Object Management Group(2016). It complements the Business ProcessModel and Notation (BPMN) with a notationfor modeling decision logic and dependenciesbetween decisions and data elements. The speci-fication formulates several goals, which can alsobe understood as hypothetical benefits: First, thenotation should be readily understandable by bothbusiness users and technical developers. Second,it should be straight forward to transform it toartifacts that implement decision logic. Third,it should be easily usable together with BPMN.DMN enjoys an increasing uptake in industry andreceives attention in academic research. However,empirical research on DMN is still scarce suchthat it is unclear to which degree the proclaimedbenefits materialize.

The aim of this paper is to structure futureresearch on DMN. Sect. 2 summarizes the back-ground of DMN. Sect. 3 describes a researchagenda for DMN, before Sect. 4 concludes thepaper.

* Corresponding author.E-mail. [email protected]

2 Background of DMN

DMN is a standard for representing operationaldecisions of day-to-day business operations. Suchdecisions are frequently taken and repetitive innature, e. g., determining if a customer is eligi-ble for an insurance cover. Common operationaldecisions often relate to the calculation and evalu-ation of business opportunities, risk managementand fraud detection. DMN complements BPMN,which does not model the decision logic in detail.DMN decouples decisions and control flow logicand it opens room for dynamic management of de-cisions. In most of the process models, decisionsare embedded within the models and scattered overprocess model constructs, eventually posing diffi-culty in maintainability (Janssens et al. 2016b). Inthis sense, DMN reduces complexity and providesa decision model which is more precise and clear(Bossuyt and Gailly 2017). In this way, DMNhelps business users in controlling their processesand organizational decisions more efficiently andeffectively by means of well-designed decisionand information structures.

More specifically, DMN defines three aspectsof decisions: the decision requirements level, thedecision logic level, and the expression language.

http://dx.doi.org/10.18417/emisa.13.2

[email protected]

International Journal of Conceptual ModelingVol. 13, No. 2 (2018). DOI:10.18417/emisa.13.2

2 Kathrin Figl, Jan Mendling, Gul Tokdemir, Jan VanthienenResearch Note

Fig. 1 illustrates these levels by the help of anexample on a seller’s credit warranting process fora potential buyer. DMN can be used together withBPMN as shown in the figure or independent ofbusiness processes.

First, the Decision Requirements Diagram(DRD) represents the relationship between de-cisions through their information requirementsand defines decision requirements through con-structs of Decision, Business Knowledge Model,Knowledge Source, Input Data, Information Re-quirement, Knowledge Requirement and Authorityrequirement (Object Management Group 2016).In Fig. 1, the input data for decision making isobtained from two sources: namely Findex andProject Specs. There is one decision whose resultis used in the business process Credit Sales andtwo intermediate decisions Credit Eligibility andCheque Performance, which yield results as inputfor the final Credit Sales decision. Payable ChequeCriteria is based on due cheque payments of thebuyer within one year with respect to project valueand project payment term. Paid Cheque Criteriaon the other hand depends on previous chequepayment history with respect to project value.

Second, the Decision Logic Level (DLL) rep-resents the logic of a single decision in the formof a boxed expression. One of the most widelyused representations for decision logic is a deci-sion table, but other expressions are allowed, e. g.using analytical models, mathematical functionsor decision rules. Decision tables define the pro-duction rules from input parameters to the outputparameters. In Fig. 1, the decision logic for CreditEligibility is shown as a decision table, in whichthe parameters Project Value, Total CredibilityAmount, Available Credit Limit, Bank Credit War-rant Letter are used as input for determining theoutput parameter Credit Eligibility.

Third, DMN also standardizes the expressionlanguage FEEL (Friendly Enough Expression Lan-guage) and a simple subset S-FEEL for use indecision tables. FEEL defines a syntax for expres-sions, which permits the description of decisionlogic in terms of decision tables, analytical mod-els, or business rules. At the bottom of Fig. 1, the

decision logic of Credit Eligibility is shown usingFEEL syntax for the rules expressed in the table.

3 Research Agenda for InvestigatingDMN Benefits

In this section, we review the literature and discussa research agenda for investigating the potentialbenefits of DMN. We structure this discussionby the help of the information systems researchframework by Hevner et al. (2004), which identi-fies technology (Sect. 3.1), individual (Sect. 3.2)and organization (Sect. 3.3) as three relevant areas.

3.1 Research Directions on TechnologicalBenefits of DMN

The history of operational decision managementand DMN finds its origin in decision table model-ing, where rules for decision logic are representedin a structure of related tables, which map com-binations of inputs to outcomes. Decision tablesand the accompanying methodology have proven apowerful vehicle for acquiring the decision knowl-edge and for checking completeness, correctnessand consistency (CODASYL Decision Table TaskGroup 1982). DMN builds upon these conceptsand goes further by standardizing existing deci-sion table formats (using a hit policy indicator),by elaborating the requirements diagram, and byintroducing a standard expression language. Eventhough DMN standardizes and extends the mod-eling capabilities of decision requirements anddecision logic (e. g. by adding FEEL), variousresults from previous research into decision tablescan be readily adopted.

Verification & Validation (V&V): Verificationand validation of rule-based systems (includingdecision tables) has been a major area of re-search, as exemplified by the earlier EUROVAVseries of conferences (European Conference onVerification & Validation of Knowledge-basedsystems) (Antoniou et al. 1998). This isimportant because at the decision logic level,decision logic is often expressed in rules andtables. There are numerous works dealingwith V&V of a set of rules (as present in



Business Process Model-BPMN

Credit Sales

Project Specifications

Credit Eligibility Rules

Credit EligibilityCheque

Performance

Paid Cheque Criteria

Paiable Cheque Criteria

Collect Application

Data

Decide on Credit Sales

Sign Contract

Decline Customer

Credit Sales=DECLINE

Credit Sales=ACCEPT

Decision Model- DMN

Decision Requirements Level

Decision Logic Level

Findex

FEEL(

decision table(

inputs:[Project Value(PV), Total Credibility Amount, Available Credit Limit, Bank,Credit Warrant Letter],

outputs:[Eligibility],

rules:[[<5K€,-,-,-,ELIGIBLE],

[ [5K€..100K€[, >=3*PV, >=PV, -, ELIGIBLE],

[ [100K€..1M€[, >=5*PV, >=1.5*PV, -, ELIGIBLE],

[ >=1M€, -,-,Available, ELIGIBLE],

[-,-,-,-,INELIGIBLE]],

hit policy: P, completeness: C))FEEL- Expression Language

Figure 1: How DMN operates solely or in relation to BPMN. An example on Credit Sales.

single decision tables). Typical rule anoma-lies are: redundancy (including duplicatesand subsumption), ambivalence, circularity,and deficiency (missing rules). Numerousalgorithms are available for checking andeliminating contradictions, redundancies andmissing rules for all possible values of theinput variables. Other approaches have beendesigned to strictly avoid table anomalies byrecommending unique single hit tables. Thus,there has been an increased interest and workin relation to verification and validation studiesin literature within the last few decades. Forexample, various tools have already includedanomaly detection algorithms (Hinkelmann2016). In a similar manner, Calvanese et al.(2016) and Laurson and Maggi (2017) proposenew algorithms to discover overlapping andmissing rules during DMN table modeling task.

In terms of correctness of decision logic, Ba-toulis et al. (2017) analyze process models withDMN by checking different soundness levelsfor decision-aware processes. For an overviewof earlier research in this area, see Vanthienenet al. (1998), or more recently Calvanese et al.(2016).

V&V over table networks: Also, V&V of tablestructures has been covered in earlier research.When input conditions or outcomes are repeatedin more than one decision, some parts of thedecision logic in a certain decision may becomeunreachable or inconsistent for specific inputvalues. Checking consistency and complete-ness between interconnected decision tables,i.e. over rule chains, is a much more challeng-ing problem than verification of single tables.




See Vanthienen et al. (1997) for an overview ofinter-tabular verification.

Table Simplification: Decision tables can besimplified and split up using normaliza-tion (Vanthienen and Dries 1994). The firsttype of simplification means that rules withequal condition inputs for all but one conditionand with the same outcomes can be joinedtogether, reducing the number of rules in thetable. This is called table contraction, minimiz-ing the number of rows for the given conditionorder. For this case, a recent study by Laursonand Maggi (2017) is a good example, where arule merging algorithm is proposed for tablesimplification. One can also identify the orderof the input conditions that leads to the mostcompact table, thereby optimizing the conditionorder. Another simplification is to split complextables into more simple ones. Decision tablescan (or should) be split up if the outcomes arenot dependent on all the conditions. This iscalled factoring or normalization, analogousto normalization in relational database theorywhere attributes should be dependent on thekey.

Code Generation: When properly specified us-ing (S)FEEL, decision models and tables areexecutable, given that appropriate DMN toolingis available. This is a straightforward executionwithout further optimization. In a number ofcases, however, attention could be paid to exe-cution efficiency or more flexible forms of codegeneration. Since the early days of decision ta-bles, a lot of work has been done in this area, bytransforming decision tables into optimal code,by generating least-cost execution trees basedon condition test times and case frequencies,see e. g. Lew (1978) or CODASYL DecisionTable Task Group (1982).

Decision Mining: Operational decisions can bemodeled in DMN by domain experts by usingthe domain knowledge present in e. g. rules, pro-cedures, policies and regulations. But decisionlogic can also be derived from case data wherethe mined model is discovered or transformed

into a decision table (Baesens et al. 2003; Wetset al. 1998). Not unlike process models, whichcan be discovered from events logs, decisiondiscovery is a form of knowledge discoveryfrom logs containing historical data about caseattributes and their outcome. Currently, deci-sion mining is often limited to discovering thedecision logic at a certain decision point in aprocess model, but a more complex challengeis the integrated mining of both a process anda decision model based on extensive decisionprocess logs (Smedt et al. 2017a).

Accordingly, literature has provided numerousresearch on how to create an integrated modelwith decisions and processes. For example, Biardet al. (2015), in their study recommend defining adecision task for multiple gateways in the processmodel and constructing a decision model as aseparated entity. They emphasize that DMN’sscope is restricted to operational level decisions,instead of tactical ones, as they are related topre-defined decisions. In the recent research,Bazhenova (2017) describes how to extract a de-cision model from process model based on splitgateways and event logs. In another study, Ba-toulis et al. (2016) bring forward an approach toadjust decision logic dynamically, using event loginformation during process implementation by cre-ating DMN model automatically that will improveprocess execution consequently. From an alterna-tive perspective, van der Aa et al. (2016) createBPMN/DMN models based on data-flow structureautomatically. Similarly, an approach for auto-matic DMN construction is defined by Bazhenovaand Weske (2015), where decision logic has beenextracted from event logs of a simple processmodel in banking domain. The research extractsdecisions from process models based on local de-cision points and limited in terms of applying theirproposal to simple process models in a specific do-main. Mertens et al. (2017) introduce DeciClare,which is a mixed-perspective process modelinglanguage. It is more for loosely framed processeswhich incorporates functional, control-flow, dataand resource views and includes concepts of DMN



language for modeling data perspective. Hence, itprovides a comprehensive view for integration ofdecisions and processes. In regard to compositionand decomposition of processes and decisions,recent research mostly focuses on simpler processmodels, where decisions are local, and do notspan across process elements other than gateways.However, complex process models, where deci-sions may extend over process modeling elementsand where dependencies exist, a holistic approachis required for integrated modeling of processesand decisions (Hasic et al. 2018). While in liter-ature the number of such research remains low,recent studies acknowledge this requirement andpropose enhanced methods. For example, Smedtet al. (2017b) put forward a holistic approach fordecision extraction from process models calledProcess Mining Integrating Decisions (P-MInD),which incorporates various business activities af-fecting the decision perspective.

DMN promises various benefits for the efficientand effective design and management of decisions,e. g. in business processes. As recent researchsuggests, decision modeling alongside businessprocesses enables and helps business users tomanage complexity (Janssens et al. 2016a; Taylor2011). This better management of complexityshould also help to support flexibility and main-tainability of processes.

Research challenges arise in this context regard-ing the consistency between DMN and BPMN.How should decisions and processes be modeledin an integrated way? How can decisions andprocesses be mined together in a combined wayto reveal entire decision structures? How canwe transform decision logic from BPMN mod-els to DMN? Answering these questions requiresresearch methods that are grounded in formalscience and design science.

3.2 Research Directions on IndividualBenefits of DMN

In order to structure the discussion of DMN-related research problems on the individual level,we refer to a theoretical model by Gemino andWand (2003) that describes modeling as a process

of knowledge construction. The outcome of thisprocess is influenced by three major perspectives:First, the characteristics of model viewers in asso-ciation with their tasks (Sect. 3.2.1); second, thecontent that is captured in the model (Sect. 3.2.2)and third, the presentation format of this content(Sect. 3.2.3). From a cognitive point of view, thecontent view relates to the inherent complexityof information that must be understood (Sweller2010). While intrinsic cognitive load may not beeasily altered without changing the decision situa-tion, extraneous cognitive load can be decreasedby how the decision model is presented and morecognitive effort can be devoted to schema con-struction (germane load) (Sweller 2010).

3.2.1 Characteristics of Model ViewersThe DMN specification (Object ManagementGroup 2016, p. 169) lists three types of possi-ble user groups: business analysts, business usersand technical developers. These user groups havedifferent technical expertise and they focus eitheron creating or on reading DMN models, respec-tively.

Novice versus Expert: Prior research on concep-tual modeling has investigated expertise fromdifferent angles. Studies including Schenk et al.(1998) have observed striking differences inthe way how novice system analysts approacha project (rather bottom-up and opportunity-driven) as compared to expert system analysts(rather top-down and goal-oriented). There arevarious requirements for a person to transitionfrom a novice to an expert status, likely also forDMN: learning the language and its concepts,developing patterns of how to capture recur-ring problems, and deliberate training over alonger period of time. The roles mentioned inObject Management Group (2016, p. 169), i. e.business analysts designing decisions, businessusers populating decision models and technicaldevelopers mapping business terms to appro-priate data technologies have different skillsand prior knowledge with respect to decisionmodeling. Also, the distinction of S-FEEL




for business users and full FEEL for advancedbusiness analysts or developers emphasizes thedifferent skill sets. Concerning domain exper-tise, decision making requires high levels ofdomain expertise (Bock 2015). Thus, in gen-eral, decisions modeled in DMN also have theadvantage to externalize such knowledge foremployees less familiar with a decision domain.

Reading and Creating Models: More gener-ally, we know from research on expertise thatbeing an expert is very much bound to a specifictask (Ericsson and Lehmann 1996). Concep-tual modeling research has distinguished theactivities of creating and reading a model. Inessence, a model viewer has to be familiar withthe syntax and semantics of a notation likeDMN in order to interpret individual models.The task of modelers is more challenging, sincethey have to transform ideas, observationsand discussions into a correct representation.Tasks of verification and validation are highlyimportant in this context, and different types ofusers might be unequally skilled for conductingthem.Besides foreseen advantages, the understand-ability of DMN by different stakeholders de-serves profound attention. In this sense, adapt-ing DMN into organizational decision makingprocesses necessitates stakeholders like busi-ness analysts or IT professionals to get famil-iar with the notation. Research could suggestmodifications to DMN to improve the notationand to shorten the learning curve of stakehold-ers in trainings. Questions arise here on howand in which circumstances users can mosteffectively work with DMN and which char-acteristics of users and models best facilitateunderstanding. Answering these questions re-quires research methods that are grounded inempirical research.

3.2.2 Semantic Content of DMNThe semantic constructs of DMN are defined bythe metamodel and have already been discussedin the previous sections. Now, we want to out-line some directions in which future research on

the cognitive difficulty of different semantic con-structs could be pursued. Similar research hasbeen conducted e. g. in the area of cognitive dif-ficulty of control flow patterns used in processmodels (Figl and Laue 2015) or of different typesof features and relations used in variability models(Reinhartz-Berger et al. 2014). If future researchis able to determine valid and reliable values forthe cognitive difficulty of understanding specificparts of decisions, such values could then be usedto guide modeling tool developers to provide feed-back on the cognitive difficulty of models to users.Model editors could calculate global metrics forthe complexity of a decision, warn the modelerwhen they exceed a certain threshold and use colorhighlighting of models and decision tables to vi-sually highlight difficult parts. Since DRGs donot depict detail information on how the decisionsare exactly taken, we deem the investigation ofthe cognitive difficulty of decision tables morepromising. For instance, future work could em-pirically assess whether unique/first hit/any hitor priority hit policies are more complex to un-derstand and lead to higher error rates. Whileautomated algorithms might check rule anoma-lies, still human comprehension of the rules isnecessary for a variety of tasks.

3.2.3 Visual Presentation of DMNWhen considering the visual presentation of DMN,we have to look separately at the three levels(DRD, DLL and FEEL). Although these levels arerelated to each other, their representation format–graphical models, tables and textual expressionlanguage–varies significantly. In the context ofthis paper, we focus on DRDs and decision ta-bles, but do not discuss the FEEL, because it ismainly textual. We structure our discussion of thepresentation of DMN into various sections basedon the physics of notations framework (Moody2009), which integrates different theoretical per-spectives to define nine principles how to designvisual notations that do not cause more cognitiveload for users than necessary. These principles



are semiotic clarity, graphic economy, visual ex-pressiveness and perceptual discriminability, se-mantic transparency, dual coding, cognitive fit,complexity management and cognitive integra-tion. Furthermore, we add one aspects whichis not explicitly defined at the notational level,but plays a critical role at the level of single dia-grams: labeling and naming conventions (Leopoldet al. 2013). We consider primary notation whichwould refer to the standard document published byOMG and relevant aspects of secondary notation,which relates to “things which are not formallypart of a notation which are nevertheless usedto interpret it” (Petre 2006, p. 293). Dangarskaet al. (2016) have presented the first analysis ofDRGs (decision tables were not evaluated) accord-ing to Moody’s framework based on an expertassessment. Besides expert evaluations, futureresearch could e. g. conduct questionnaire-basedstudies to provide user evaluations of the sym-bol set of DMN, e. g. by using scales to assesssemiotic clarity, perceptual discriminability, vi-sual expressiveness or semantic transparency (seefor instance evaluations of symbol sets of othermodeling notations (Figl et al. 2010, 2013)). An-other approach could be to develop symbol setsoptimized based on cognitive design guidelines,which has been done for other existing notations(Genon et al. 2012, 2011) and compare their effecton comprehensibility with original DMN sym-bols. Moreover, usability tests and experimentsincluding eye-tracking could be performed by re-searchers to assess the understandability of thevisual notation of DMN. The following sectionsgives an background on relevant factors whichcould be included in experiments and highlightsrelevant open research questions.

Semiotic ClarityThe principle of semiotic clarity demands thatthere is a 1:1 relationship between any semanticconstruct and the corresponding visual symbol thenotation offers. One potential violation of this rulein the form of symbol redundancy (more than onevisual representation for one and the same underly-ing semantic construct) can for instance be found

in the DMN notation (Object Management Group2016, p. 31): “An alternative compliant way to dis-play requirements for Input Data, especially usefulwhen DRDs are large or complex, is that InputData are not drawn as separate notational elementsin the DRD, but are instead listed on those Deci-sion elements which require them.” In the contextof semantic constructs, a representational analysisaccording to the theory by Wand/Weber based onBunge (Recker et al. 2011) may also be effective tohighlight the concepts which DMN is supportingin contrast to other decision modeling notations.An interesting starting point for such an analysismight be a paper by Bock (2015), who has startedto identify and compare semantic constructs fordecision making in various visual modeling ap-proaches, e. g. decision matrices, decision treesand influence diagrams. In comparison to otherapproaches, DMN has a strong focus on routineoperational decisions and is less suitable to am-biguous, non-routine and novel decision situations(Bock 2015).

Graphic EconomyIn comparison to other modeling notations likeBPMN, which offer a high number of symbols,DMN can be considered parsimonious because thesize of its vocabulary it manageable (4 symbols,3 types of edges for different requirements and avisual definition for using textual annotations forDRDs (Object Management Group 2016, p. 30)).Based on a theoretical approach analyzing theobjects, relationships and properties of the meta-model, the cumulative complexity of DMN wasrated “relatively low”, comparable to the modelingstandard CMMN, but lower than BPMN, whichoffers higher expressive power (Hasic et al. 2017).In the authors’ assessment they conclude that“DMN should be simple to learn and understand”(Hasic et al. 2017, p. 69).

Visual Expressiveness and PerceptualDiscriminabilityThe use of visual variables for symbols (position,color, size, texture, shape, orientation, bright-ness) determines the visual expressiveness of a




notation. Pairwise differences of symbols on vi-sual variables increase perceptual discriminability.In general, DMN uses rectangles to representdecisions and variations of rectangles for otherconcepts: e. g. for the concept Knowledge Sourcea “shape with three straight sides and one wavyone”, for Business Knowledge Model a “rectanglewith two clipped corners” and for Input Data a“shape with two parallel straight sides and twosemi-circular ends” (Object Management Group2016, p. 31). Requirements differ according tothe texture (dotted vs. solid lines) and type ofarrowhead. Therefore, Dangarska et al. (2016)have rated visual expressiveness of DMN as lowand perceptual discriminability as partly violated.Empirical research is necessary to find out whetherlow visual expressiveness and discriminability ac-tually lead to problems for users to distinguishsymbols.

Semantic TransparencySemantic transparency is determined by how eas-ily the meaning of a visual appearance can be“inferred from its appearance ” (Moody 2009,p. 764). Dangarska et al. (2016) have giventhe core symbols an opaque score (indicatingan “arbitrary relationship between appearance andmeaning”), while arrows for Requirements wererated as immediate. Semantic transparency isalso relevant for choosing spatial arrangementsof elements that ease the comprehension of theirrelationship. The DMN standard does not giveconcrete instructions on how to arrange and layoutDRDs. When looking at simple DRDs drawnin the standard document (Object ManagementGroup 2016, p. 75), input data and sub-decisionsare placed below decisions, business knowledgesymbols are placed left and knowledge sourcessymbols are placed on the left and on the rightabove decisions. Empirical research could testwhether placement of model elements has any ef-fect on the comprehension of DRDs and whetherthere are positioning guidelines of elements whichwould result in easier to understand DRDs. Instudies on interpreting diagrams with nonsensewords “causes were always thought to lie to the

left of and above effects” (Winn 1990, p. 155).Thus, placing input data and sub-decisions tothe left or above decisions might also be intu-itively understandable. However, tree structuresexpanding from a parent node at the top or fromthe left are also widely-used conventions, sup-porting the exemplary spatial arrangement in thestandard document (Object Management Group2016, p. 75). Concerning the layout of decisiontables the DMN standard document also givesusers freedom of choice and states “a decisiontable can be presented horizontally (rules as rows),vertically (rules as columns), or crosstab (rulescomposed from two input dimensions)” (ObjectManagement Group 2016, p. 75). However, thestandard is quite clear on the positioning of in-put columns, which reflect the findings of Winn(1990) on intuitively understandable conventionsof spatial arrangements: “In a horizontal table,all input columns SHALL be represented on theleft of all output columns. In a vertical table, allthe input rows SHALL be represented above alloutput rows” (Object Management Group 2016,p. 75).

Dual CodingWhile text should not be used to distinguish be-tween symbols, it can be wise to supplement graph-ical information with it (Moody 2009). DMN ac-tively encourages text annotations, using a dottedline and a square bracket. Furthermore, it givesclear advice on how to combine text, e. g. “thelabel . . . SHALL be clearly inside the shape ofthe DRD element” (Object Management Group2016, p. 31). Such a guideline can theoreticallygrounded on the Gestalt law of common region,which posits “the tendency for elements that liewithin the same bounded area to be grouped to-gether” (Palmer et al. 2003, p. 312). However,user evaluations of business process models haveshown that readers rate labels placed physicallyclose to symbols equally well to labels placedinside a symbol (Figl 2017).



Cognitive FitMoody (2009) suggests the use of different visualdialects for experts and novices as well as for dif-ferent representational media in order to achievecognitive fit. This is something DMN does notoffer. However, DMN addresses the issue of cog-nitive fit in the wider sense since a main objectiveof DMN is to combine decision tables and re-quirements diagrams (DRDs) to account for thefact that both are well suited to represent differenttypes of information elements for different tasks.Cognitive fit theory (Vessey and Galletta 1991)originates from the observation that graph versustable use is suited for different tasks. Still, it mightbe advantageous to offer users not only decisiontables, but also decision trees as additional visual-ization to comprehend a complex decision. Vesseyand Weber (1986) compared decision tables anddecision trees in the context of a programming taskand found decision trees to outperform decisiontables. Similarly, decision trees were found to bemore helpful when used in a investment game thanthe corresponding decision tables (Subramanianet al. 1992). However, for various comprehensiontasks more recent experiments revealed that deci-sion tables performed better than decision treesand textual rules (Huysmans et al. 2011). Overall,more work is needed to clarify inconsistenciesof prior research and address whether offeringdecision tree visualizations in addition to decisiontables as specified in DMN might enhance humancomprehension.

Complexity Management and HierarchicalStructuring of DecisionsTo avoid overloading human working memorywith large and complex diagrams, Moody (2009,p. 766) suggests that visual notations should pro-vide mechanisms for modularization and hier-archically structuring. Hierarchical structuringof decisions and modularity are a main purposeof DRG. DMN allows the modeler to split updecisions into different tables and specify theirconnection. DMN leaves it to the implementationsof the modeling tool to show interactive visualiza-tions of Decision Requirements Graphs (DRG) in

an efficient way: “For any significant domain ofdecision-making a DRD representing the completeDRG may be a large and complex diagram. Imple-mentations MAY provide facilities for displayingDRDs which are partial or filtered views of theDRG, e. g., by hiding categories of elements, orhiding or collapsing areas of the network. DMNdoes not specify how such views should be notated,but whenever information is hidden implementa-tions SHOULD provide a clear visual indicationthat this is the case” (Object Management Group2016, p. 35). Decomposition of decisions (split-ting decisions into sub-decisions) is relevant forDRDs and well as decision tables, as the numberof input variables is visually shown in the DRDsas well as part of the decision tables. Mertenset al. (2015, p. 161) note “The decision table rep-resentation also has some drawbacks. When thedecisions themselves are based on a very largeamount of conditions and actions, the readabilityof a table gets lost. In such cases, the decisiontable will need to be split in multiple smaller ta-bles to allow them to stay manageable.” Thereis a long history of literature on decision tabledesign, offering guidance to structure decisionsinto separate tables, to build decision tables usinga stepwise methodology and to avoid table anoma-lies and unnormalized tables. For an overviewof guidelines for decision table, see e. g. CODA-SYL Decision Table Task Group (1982). A recenttutorial on using DMN by Signavio for instancesuggests to split decisions into sub-decisions assoon as a decision has 7 or more inputs. Overall,decomposition in models can lead to two differenteffects: despite the positive effect of abstraction,which eases comprehension and has lead to thecommon belief that hierarchically structuring isbeneficial for model comprehension, it can alsoleads to a split-attention effect as readers have toswitch between different models or tables (Zugal etal. 2012). Depending on the comprehension task,fully flattened models, respectively decision ta-bles including all input variables may even lead tohigher comprehension, as experiments have shownin other modeling domains, e. g. process models(Turetken et al. 2016) or data models (Parsons




2002). While the problem is not new, research forthe specific nature of decision modeling is needed.The effects of hierarchical structuring will alsodiffer according to the complexity managementand cognitive integration mechanism offered bythe modeling tools used. When using table simpli-fication, contraction and normalization, it is alsoimportant to consider their effect on repeated useand interactivity of elements; elements that heavilyinteract cannot be comprehended in isolation andheighten the cognitive load of understanding thedecision tables (Sweller 2010). Prior research hasdemonstrated that different modeling strategiesrelated to minimality and repeated use of elements(e. g. for structuring features in feature trees andusing cross-cutting concerns (Reinhartz-Bergeret al. 2017)) highly affect model comprehension.

Cognitive IntegrationBoth, homogeneous and heterogeneous integra-tion are relevant for DMN: heterogeneous integra-tion, because it is important for users to understandthe bridge DRDs form to different types of visualrepresentations (especially BPMN diagrams anddecision logic tables); homogeneous integration,because more than one diagram of the same type(DRD) can depict a DRG. Although the visuali-sation of DRDs should lower the risk of hiddendependencies, which are relationship betweencomponents “such that one of them is dependenton the other, but that the dependency is not fullyvisible” (Green and Petre 1996, p. 153), under-standing interconnected decision tables might gethard and should be investigated in empirical stud-ies.

Labeling and Naming ConventionsLabels carry the semantic content. Modelingsymbols as decisions, input data or knowledgesource as well as input and output columns ofthe decision tables have to be labeled. For labelsof process models, research has already demon-strated that users rate verb-object label styles (e. g.“Determine discount”) for tasks as most useful andleast ambiguous (Mendling et al. 2010). Sincethe top-level decision corresponds to a businessrule tasks in a BPMN diagram, a DMN tutorial

(Signavio 2017) recommends to use exactly thesame label (in verb-object label style). There areother labeling/naming styles as output style (“Dis-count”) and question style (“Does the customerget a discount?”) and Signavio (2017) give thefollowing advice: “It is best to use the outputstyle for all other decisions, but in some cases thequestion style is more intuitive than the outputstyle.” If labels get longer as it is the case in thequestion style, segmentation and visual designof labels gets more critical (Koschmider et al.2016). However, empirical research testing theactual effects of naming conventions and visualdesign of labels are still missing for process mod-els, thus their results cannot be directly transferredto DMN and the specifics of DMN demand aseparate empirical evaluation anyway. Addressingthis challenge requires future empirical researchbuilding on experimental designs.

3.3 Research Directions onOrganizational Benefits of DMN

Decisions that are explicitly defined through DMNand not hardcoded inside organizational decisionmaking processes will likely decrease complexityand hence ease the implementation of businessrules and analytic technologies. In this way, DMNmight contribute to improved efficiency and effec-tiveness of organizational decision making, e. g.in terms of increased agility (Jonkers et al. 2013),improved business/IT alignment and increasedstraight-through processing (Taylor et al. 2013).From an organizational point of view, Lemmens(2015) also emphasize the importance of integra-tion of modeling notions, that organizations utilizeand develop, namely of process modeling, infor-mation modeling and rules modeling, for theirorganizational goals and operations for agility.

From another perspective, it is also clear thatdecision execution efficiency is highly affectedby the amount of input data that is required to becollected for business process decisions, whichis likewise costly for organizations. However, arecent study by Bazhenova and Weske (2017) pro-poses a method to reduce cost of the input data



collection process through a prioritization algo-rithm, which might lead to further organizationalbenefits.

It is also suggested that employing DMN in or-ganizational decision making might be specificallyuseful in a setting where business rules changefrequently and where decisions have high riskfor the operations. Hence, sectors like financialservices, insurance, energy and ICT providers arelisted as recommended areas of use. So far, vari-ous processes with strong regulatory requirementsin the financial industry are currently redesignedand formalized using DMN. Among others, theseinclude the know-your-customer process, which issubject to regulations for anti-money launderingand counter-terrorist financing rules. Other sec-tors like health care (Combi et al. 2016; Servadeiet al. 2017; Wiemuth et al. 2017), disaster man-agement (Horita et al. 2016), retailers or logisticsmight benefit in a similar fashion.

On the other hand, as DMN allows separationof control flow and decisions, it may also providecooperation and can be shared between differentstakeholders in Collaborative Networks (Biard etal. 2015) and increase solidly the level of benefi-ciary gains in organizational goals, settings andoutcomes, all of which extends and broadens thelimits of research within this domain.

DMN besides, is taking a role in informationsystem development field as well. In the recent ex-ample of Boumahdi et al. (2016), DMN is utilizedfor defining decision view of the service designin SOA, which may be extended to create ModelDriven Architectures automatically. In a similarway, it is also emphasized that the use of DMN isconsiderable with other conceptual models usedin information system development phases likerequirement specification (Kluza et al. 2017) toimprove decision logic definition. Thus, benefitsof DMN, in parallel to organizational ones, insystem development field and how to incorporatein various phases is another area of research to beexplored.

Questions on this organizational level relateto how and in which circumstances, companies

adopting DMN can realize these proclaimed ben-efits and what success factors play an importantrole here. Answering these questions requiresfurther research with diverse empirical researchmethods and cases in this regard.

4 Conclusion

DMN will change the way how processes arespecified and implemented. In this paper, wedescribed its technical foundations of decisiontable research and its theoretical background ofmodeling research. We identified research direc-tions for investigating its potential benefits on atechnological, individual and organizational level,and in this way clarifying what we know and whatwe don’t know about DMN. Insights into the wayhow programmed decisions are specified and im-plemented together with business process will bea cornerstone of future research into informationsystems and business process management in theyears to come.

References

Antoniou G., van Harmelen F., Plant R., Van-thienen J. (1998) Verification and validation ofknowledge-based systems: Report on two 1997events. In: AI Magazine 19(3), pp. 123–126

Baesens B., Setiono R., Mues C., VanthienenJ. (2003) Using neural network rule extractionand decision tables for credit-risk evaluation. In:Management science 49(3), pp. 312–329

Batoulis K., Baumgraß A., Herzberg N., Weske M.(2016) Enabling Dynamic Decision Making inBusiness Processes with DMN. In: Reichert M.,Reijers H. A. (eds.) Business Process ManagementWorkshops 2015. Lecture Notes in Business In-formation Processing Vol. 256. Springer, pp. 418–431

Batoulis K., Haarmann S., Weske M. (2017) Var-ious Notions of Soundness for Decision-AwareBusiness Processes. In: ER. Lecture Notes in Com-puter Science Vol. 10650. Springer, pp. 403–418




Bazhenova E. (2017) Optimization of DecisionMaking in Business Processes. In: Proceedingsof the 9th Ph. D. retreat of the HPI ResearchSchool on service-oriented systems engineering100, pp. 9–19

Bazhenova E., Weske M. (2015) Deriving decisionmodels from process models by enhanced decisionmining. In: International Conference on BusinessProcess Management. Springer, pp. 444–457

Bazhenova E., Weske M. (2017) Optimal acquisi-tion of input data for decision taking in businessprocesses. In: Proceedings of the Symposium onApplied Computing. ACM, pp. 703–710

Biard T., Mauff A. L., Bigand M., Bourey J. P.(2015) Separation of Decision Modeling fromBusiness Process Modeling Using New "Deci-sion Model and Notation" (DMN) for Automat-ing Operational Decision-Making. In: PRO-VE.IFIP Advances in Information and Communica-tion Technology Vol. 463. Springer, pp. 489–496

Bock A. (2015) The Concepts of Decision Making:An Analysis of Classical Approaches and Avenuesfor the Field of Enterprise Modeling In: PoEMRalyté J., España S., Pastor Ó. (eds.) Vol. 235Lecture Notes in Business Information ProcessingSpringer, pp. 306–321

Bossuyt J., Gailly F. (2017) investigating the ben-efits of modeling business processes with BPMN+ DMN. MA thesis, University of Gent, Belgium

Boumahdi F., Chalal R., Guendouz A., Gasmia K.(2016) SOA+d: a new way to design the decisionin SOA—based on the new standard DecisionModel and Notation (DMN). In: Service OrientedComputing and Applications 10(1), pp. 35–53

Calvanese D., Dumas M., Laurson Ü., MaggiF. M., Montali M., Teinemaa I. (2016) Semanticsand Analysis of DMN Decision Tables. In: RosaM. L., Loos P., Pastor O. (eds.) Business ProcessManagement 2016, Proceedings. Lecture Notes inComputer Science Vol. 9850. Springer, pp. 217–233

CODASYL Decision Table Task Group (1982) Amodern appraisal of decision tables. Associationfor Computing Machinery

Combi C., Oliboni B., Zardiniy A., Zerbato F.(2016) Seamless Design of Decision-IntensiveCare Pathways. In: 2016 IEEE International Con-ference on Healthcare Informatics, ICHI 2016,Chicago, IL, USA, October 4-7, 2016. IEEE Com-puter Society, pp. 35–45

Dangarska Z., Figl K., Mendling J. (2016) AnExplorative Analysis of the Notational Character-istics of the Decision Model and Notation (DMN).In: Dijkman R. M., Pires L. F., Rinderle–Ma S.(eds.) EDOC Workshops 2016. IEEE ComputerSociety, pp. 1–9

Ericsson K. A., Lehmann A. C. (1996) Expert andexceptional performance: Evidence of maximaladaptation to task constraints. In: Annual reviewof psychology 47(1), pp. 273–305

Figl K. (2017) User Evaluation of Symbols forCore Business Process Modeling Concepts. In:Ramos I., Tuunainen V., Krcmar H. (eds.) 25thEuropean Conference on Information Systems,ECIS 2017, Guimarães, Portugal, June 5-10, 2017,pp. 581–594

Figl K., Derntl M., Rodriguez M. C., Botturi L.(2010) Cognitive Effectiveness of Visual Instruc-tional Design Languages. In: Journal of VisualLanguages & Computing 21(6), pp. 359–373

Figl K., Laue R. (2015) Influence Factors forLocal Comprehensibility of Process Models. In:International Journal of Human-Computer Studies82, pp. 96–110

Figl K., Recker J., Mendling J. (2013) A studyon the effects of routing symbol design on pro-cess model comprehension. In: Decision SupportSystems 54(2), pp. 1104–1118

Gemino A., Wand Y. (2003) Evaluating model-ing techniques based on models of learning. In:Communications of the ACM 46(10), pp. 79–84



Genon N., Caire P., Toussaint H., Heymans P.,Moody D. L. (2012) Towards a More Semanti-cally Transparent i* Visual Syntax. In: Regnell B.,Damian D. E. (eds.) Requirements Engineering:Foundation for Software Quality - 18th Interna-tional Working Conference, REFSQ 2012, Es-sen, Germany, March 19-22, 2012. Proceedings.Lecture Notes in Computer Science Vol. 7195.Springer, pp. 140–146

Genon N., Heymans P., Amyot D. (2011)Analysing the Cognitive Effectivenessof the BPMN 2.0 Visual Notation. In:Malloy B. A., Staab S., van den Brand M.(eds.) Software Language Engineering - ThirdInternational Conference, SLE 2010, Eindhoven,The Netherlands, October 12-13, 2010, RevisedSelected Papers. Lecture Notes in ComputerScience Vol. 6563. Springer, pp. 377–396

Green T., Petre M. (1996) Usability Analysis ofVisual Programming Environments: A ’CognitiveDimensions’ Framework. In: Journal of VisualLanguages & Computing 7(2), pp. 131–174

Hasic F., De Smedt J., Vanthienen J. (2017) To-wards Assessing the Theoretical Complexity of theDecision Model and Notation (DMN). In: GuldenJ., Nurcan S., Reinhartz-Berger I., Guedria W.,Bera P., Guerreiro S., Fellmann M., Weidlich M.(eds.) Joint Proceedings of the Radar tracks atthe 18th International Working Conference onBusiness Process Modeling, Development andSupport (BPMDS), and the 22nd InternationalWorking Conference on Evaluation and ModelingMethods for Systems Analysis and Development(EMMSAD), and the 8th International Workshopon Enterprise Modeling and Information SystemsArchitectures (EMISA). CEUR-WS.org Vol. 1859,pp. 64–71

Hasic F., Devadder L., Dochez M., Hanot J., SmedtJ. D., Vanthienen J. (2018) Challenges in Refac-toring Processes to Include Decision Modelling.In: Teniente E., Weidlich M. (eds.) Business Pro-cess Management Workshops - BPM 2017 Inter-national Workshops, Barcelona, Spain, Septem-ber 10-11, 2017, Revised Papers. Lecture Notes

in Business Information Processing Vol. 308.Springer, pp. 529–541

Hevner A. R., March S. T., Park J., Ram S. (2004)Design science in information systems research.In: MIS quarterly 28(1), pp. 75–105

Hinkelmann K. (2016) Business Process Flexibil-ity and Decision-Aware Modeling - The Knowl-edge Work Designer. In: Karagiannis D., MayrH. C., Mylopoulos J. (eds.) Domain-Specific Con-ceptual Modeling, Concepts, Methods and Tools.Springer, pp. 397–414

Horita F. E. A., Link D., de Albuquerque J. P.,Hellingrath B. (2016) oDMN: An IntegratedModel to Connect Decision-Making Needs toEmerging Data Sources in Disaster Management.In: Bui T. X., Jr. R. H. S. (eds.) 49th Hawaii Inter-national Conference on System Sciences, HICSS2016, Koloa, HI, USA, January 5-8, 2016. IEEEComputer Society, pp. 2882–2891

Huysmans J., Dejaeger K., Mues C., Vanthienen J.,Baesens B. (2011) An empirical evaluation of thecomprehensibility of decision table, tree and rulebased predictive models. In: Decision SupportSystems 51(1), pp. 141–154

Janssens L., Bazhenova E., De Smedt J., Van-thienen J., Denecker M. (2016a) Consistent Inte-gration of Decision (DMN) and Process (BPMN)Models. In: España S., Ivanovic M., Savic M.(eds.) Proceedings of the CAiSE’16 Forum. CEURWorkshop Proceedings Vol. 1612, pp. 121–128

Janssens L., Smedt J. D., Vanthienen J. (2016b)Modeling and Enacting Enterprise Decisions. In:Krogstie J., Mouratidis H., Su J. (eds.) AdvancedInformation Systems Engineering Workshops -CAiSE 2016 International Workshops, Ljubljana,Slovenia, June 13-17, 2016, Proceedings. Lec-ture Notes in Business Information ProcessingVol. 249. Springer, pp. 169–180

Jonkers H., Quartel D., van den Berg H., FrankenH. (2013) Integration of Business Decision Mod-eling in Organization Design




Kluza K., Wisniewski P., Jobczyk K., Ligeza A.,Suchenia A. (2017) Comparison of Selected Mod-eling Notations for Process, Decision and SystemModeling. In: Ganzha M., Maciaszek L. A., Pa-przycki M. (eds.), pp. 1095–1098

Koschmider A., Figl K., Schoknecht A. (2016)A Comprehensive Overview of Visual Design ofProcess Model Element Labels. In: Reichert M.,Reijers H. A. (eds.) Business Process ManagementWorkshops - BPM 2015, 13th International Work-shops, Innsbruck, Austria, August 31 - September3, 2015, Revised Papers. Lecture Notes in Busi-ness Information Processing Vol. 256. Springer,pp. 571–582

Laurson Ü., Maggi F. M. (2017) A Tool for theAnalysis of DMN Decision Tables. In: AzevedoL., Cabanillas C. (eds.) Proceedings of the BPMDemo Track 2016 Co-located with the 14th In-ternational Conference on Business Process Man-agement (BPM 2016), Rio de Janeiro, Brazil,September 21, 2016.. CEUR Workshop Proceed-ings Vol. 1789. CEUR-WS.org, pp. 56–60

Lemmens I. M. C. (2015) Integrating ModellingDisciplines at Conceptual Semantic Level. In:Ciuciu I., Panetto H., Debruyne C., Aubry A.,Bollen P., Valencia-García R., Mishra A., FenselA., Ferri F. (eds.) On the Move to Meaning-ful Internet Systems: OTM 2015 Workshops- Confederated International Workshops: OTMAcademy, OTM Industry Case Studies Program,EI2N, FBM, INBAST, ISDE, META4eS, andMSC 2015, Rhodes, Greece, October 26-30, 2015,Proceedings. Lecture Notes in Computer ScienceVol. 9416. Springer, pp. 188–196

Leopold H., Eid-Sabbagh R., Mendling J.,Azevedo L. G., Baião F. A. (2013) Detectionof naming convention violations in process mod-els for different languages. In: Decision SupportSystems 56, pp. 310–325

Lew A. (1978) Optimal conversion of extended-entry decision tables with general cost criteria. In:Communications of the ACM 21(4), pp. 269–279

Mendling J., Reijers H. A., Recker J. (2010) Ac-tivity Labeling in Process Modeling: EmpiricalInsights and Recommendations. In: InformationSystems 35(4), pp. 467–482

Mertens S., Gailly F., Poels G. (2015) EnhancingDeclarative Process Models with DMN DecisionLogic In: Enterprise, Business-Process and Infor-mation Systems Modeling: 16th International Con-ference, BPMDS 2015, 20th International Con-ference, EMMSAD 2015, Held at CAiSE 2015,Stockholm, Sweden, June 8-9, 2015, ProceedingsGaaloul K., Schmidt R., Nurcan S., Guerreiro S.,Ma Q. (eds.) Springer International Publishing,Cham, pp. 151–165

Mertens S., Gailly F., Poels G. (2017) Towardsa decision-aware declarative process modelinglanguage for knowledge-intensive processes. In:Expert Systems with Applications 87, pp. 316–334

Moody D. L. (2009) The Physics of Notations:Toward a Scientific Basis for Constructing VisualNotations in Software Engineering. In: IEEE Trans.Software Eng. 35(6), pp. 756–779

Object Management Group (2016) DecisionModel and Notation (DMN), Version 1.0. Nor-mative Document

Palmer S. E., Brooks J. L., Nelson R. (2003) Whendoes grouping happen? In: Acta Psychologica114(3), pp. 311–330

Parsons J. (2002) Effects of Local Versus GlobalSchema Diagrams on Verification and Communi-cation in Conceptual Data Modeling. In: Journal ofManagement Information Systems 19(3), pp. 155–183

Petre M. (2006) Cognitive Dimensions "Beyondthe Notation". In: Journal of Visual Languages &Computing 17(4), pp. 292–301

Recker J., Rosemann M., Green P. F., IndulskaM. (2011) Do Ontological Deficiencies in Model-ing Grammars Matter? In: MIS Quarterly 35(1),pp. 57–79



Reinhartz-Berger I., Figl K., Haugen Ø. (2014)Comprehending Feature Models Expressed inCVL. In: Dingel J., Schulte W., Ramos I., AbrahãoS., Insfran E. (eds.) International Conference onModel-Driven Engineering Languages and Sys-tems (MODELS). Springer International Publish-ing, pp. 501–517

Reinhartz-Berger I., Figl K., Haugen Ø. (2017)Investigating Styles in Variability Modeling: Hi-erarchical vs. Constrained Styles. In: Informationand Software Technology 87, pp. 81–102

Schenk K. D., Vitalari N. P., Davis K. S. (1998)Differences between novice and expert systemsanalysts: What do we know and what do we do?In: Journal of Management Information Systems15(1), pp. 9–50

Servadei L., Schmidt R., Bär F. (2017) ArtificialNeural Network for Supporting Medical Deci-sion Making: A Decision Model and NotationApproach to Spondylolisthesis and Disk Hernia.In: Ciuciu I., Debruyne C., Panetto H., WeichhartG., Bollen P., Fensel A., Vidal M. (eds.) On theMove to Meaningful Internet Systems: OTM 2016Workshops - Confederated International Work-shops: EI2N, FBM, ICSP, Meta4eS, and OTMA2016, Rhodes, Greece, October 24-28, 2016, Re-vised Selected Papers. Lecture Notes in ComputerScience Vol. 10034. Springer, pp. 217–227

Signavio (2017). https://mooc.house/courses/signavio-%7BDMN%7D-intro

Smedt J. D., vanden Broucke S. K. L. M., ObregonJ., Kim A., Jung J., Vanthienen J. (2017a) DecisionMining in a Broader Context: An Overview of theCurrent Landscape and Future Directions. In: Du-mas M., Fantinato M. (eds.) Business Process Man-agement Workshops - BPM 2016 InternationalWorkshops, Rio de Janeiro, Brazil, September 19,2016, Revised Papers. Lecture Notes in BusinessInformation Processing Vol. 281, pp. 197–207

Smedt J. D., Hasic F., vanden Broucke S. K. L. M.,Vanthienen J. (2017b) Towards a Holistic Discov-ery of Decisions in Process-Aware InformationSystems. In: Carmona J., Engels G., Kumar A.

(eds.) Business Process Management - 15th Inter-national Conference, BPM 2017, Barcelona, Spain,September 10-15, 2017, Proceedings. LectureNotes in Computer Science Vol. 10445. Springer,pp. 183–199

Subramanian G. H., Nosek J., Raghunathan S. P.,Kanitkar S. S. (1992) A comparison of the decisiontable and tree. In: Communications of the ACM35(1), pp. 89–94

Sweller J. (2010) Element Interactivity and Intrin-sic, Extraneous, and Germane Cognitive Load. In:Educational Psychology Review 22(2), pp. 123–138

Taylor J. (2011) Decision management systems: apractical guide to using business rules and predic-tive analytics. Pearson Education

Taylor J., Fish A., Vanthienen J., Vincent P. (2013)Emerging Standards in Decision Modeling: AnIntroduction to Decision Model & Notation. In:iBPMS. Strategies Inc., pp. 133–146

Turetken O., Rompen T., Vanderfeesten I., DikiciA., van Moll J. (2016) The effect of modularityrepresentation and presentation medium on theunderstandability of business process models inBPMN. In: International Conference on BusinessProcess Management. Springer, pp. 289–307

van der Aa H., Leopold H., Batoulis K., WeskeM., Reijers H. A. (2016) Integrated Process andDecision Modeling for Data-Driven Processes. In:Reichert M., Reijers H. A. (eds.) Business Pro-cess Management Workshops 2015. Lecture Notesin Business Information Processing Vol. 256.Springer, pp. 405–417

Vanthienen J., Dries E. (1994) Illustration of a de-cision table tool for specifying and implementingknowledge based systems. In: International Jour-nal on Artificial Intelligence Tools 3(02), pp. 267–288

Vanthienen J., Mues C., Aerts A. (1998) An il-lustration of verification and validation in themodelling phase of KBS development. In: Data &Knowledge Engineering 27(3), pp. 337–352


https://mooc.house/courses/signavio-%7BDMN%7D-intro

https://mooc.house/courses/signavio-%7BDMN%7D-intro



Vanthienen J., Mues C., Wets G. (1997) Inter-Tabular Verification in an Interactive Environment.In: Vanthienen J., van Harmelen F. (eds.) Pro-ceedings of the Fourth European Symposium onthe Validation and Verification of Knowledge-Based Systems, EUROVAV’97, June 26-28, 1997,Katholieke Universiteit Leuven, Leuven, Belgium.Katholieke Universiteit Leuven, Belgium, pp. 155–165Vessey I., Galletta D. F. (1991) Cognitive Fit: AnEmpirical Study of Information Acquisition. In:Information Systems Research 2(1), pp. 63–84Vessey I., Weber R. (1986) Structured tools andconditional logic: an empirical investigation. In:Commun. ACM 29(1), pp. 48–57Wets G., Vanthienen J., Timmermans H. J. P.(1998) Modelling Decision Tables from Data. In:Wu X., Ramamohanarao K., Korb K. B. (eds.) Re-search and Development in Knowledge Discoveryand Data Mining, Second Pacific-Asia Conference,PAKDD-98, Melbourne, Australia, April 15-17,1998, Proceedings. Lecture Notes in ComputerScience Vol. 1394. Springer, pp. 412–413Wiemuth M., Junger D., Leitritz M., NeumannJ., Neumuth T., Burgert O. (2017) Applicationfields for the new Object Management Group(OMG) Standards Case Management Model andNotation (CMMN) and Decision ManagementNotation (DMN) in the perioperative field. In:International Journal of Computer Assisted Radi-ology and Surgery, pp. 1–11Winn W. (1990) Encoding and retrieval of informa-tion in maps and diagrams. In: IEEE Transactionson Professional Communication 33(3), pp. 103–107Zugal S., Pinggera J., Weber B., Mendling J.,Reijers H. A. (2012) Assessing the Impact ofHierarchy on Model Understandability - A Cog-nitive Perspective. In: Kienzle J. (ed.) Models inSoftware Engineering - Workshops and Symposiaat MODELS 2011, Wellington, New Zealand,October 16-21, 2011, Reports and Revised Se-lected Papers. Lecture Notes in Computer ScienceVol. 7167. Springer, pp. 123–133

This work is licensed undera Creative Commons“Attribution-ShareAlike 4.0International” license.


https://creativecommons.org/licenses/by-sa/4.0/deed.en




Documents

WhatweknowandwhatwedonotknowaboutDMNkathrinfigl.com/wp-content/uploads/2018/04/What-we...Business Process Model-BPMN Credit Sales Project Specifications Credit Eligibility Rules Credit