An Empirical Study on Simplification of Business Process Modeling Languages (BPMN). Presentation at SLE2015 & SPLASH2015 Pittsburg, PA

An Empirical Study on Simplification ofBusiness Process Modeling Languages

Marco Brambilla, Eric Umuhoza, Politecnico di Milano, ItalyJordi Cabot, ICREA & UOC, SpainDavide Ripamonti, Fluxedo, Italy

Agenda

• Motivation• Problem setting• Process• Experience reporting• Success story (?)

Context and Motivation

• Adaptation of modeling languages Standard languages are complex No perfect match of the domain to be modeled

• Other approaches towards simplification New DSLs Extending an existing base language

• Our approach Simplify existing language according to the user needs

Empirical experiment: BPM scenario

Michael zur Muehlen and Jan Recker "How much language is enough? "

How much is enough?

Objective: Personal Processes

From BPM to PPM

Study “how much is enough” for• End users• Collaborative planning and execution• Social network based interactions

Meanwhile on the TODOlist planet…

Commercial Question

Any Space for intermediate solutions?

Our Study

Our ProcessEn

d-us

erLa

ngua

ge d

esig

ner

UserQuestionnaire

Language Evaluation

Definition of Language Variants

Modeling of Use Cases

Simplified LanguageSelection of

Reducible Language Elements

General Language

Beyond classical approaches of language quality

(e.g., Moody’s physics)

1. Semiotic Clarity 2. Perceptual Discriminability 3. Semantic Transparency 4. Complexity Management5. Cognitive Integration 6. Visual Expressiveness 7. Dual Coding 8. Graphic Economy 9. Cognitive Fit

Not just about the Syntax

Before going to syntax, you need to address semantics!

• Identify possible reduction points• Select variations of those points• Cluster them (too many combinations!)

Selection of reducible elements

Syntax VariantsElements to evaluate Syntax 1 Syntax 2 Syntax 3 Syntax 4

Start x x x xEnd x x x xTask x x x xParams: global x xParams: single local x

Params: multiple local x

Events x x xSequence x x x xParallel x x xCondition x x xCycle x

Implementation details

Online model editor (PHP, HTML5, CSS3, JS JQuery)• Maximum usability• Configurable for syntax variants• Tracking user activity• Minimal model checking realitme

Validation

Experiment setup• 3 application scenarios• 4 syntax variants• 24 users• Multiple tests per user

Variants Assigment

4 syntaxes3 scenarios

Graeco-latin square:• 2 cases per user• No replication of sytax nor

scenario

} = 12 cases

Variants Assigment

4 syntaxes3 scenarios

Graeco-latin square:• The real one

Procedure with users

1. Intro2. Learning3. Experiment

1. Learn syntax 2. Read scenario3. Model scenario with syntax

4. Questionnaire 1. Demographics 2. Evaluation of experience

Results Analysis

• Average modeling time • Average # of used concepts

~16 min

~21 min

~19 min

Language VariantsLanguage Variants

Dur

atio

n (s

)

# E

lem

ents

Modeling operations – average count

Results Analysis

Wait Until Parallel Condition Cycle Activity Sequence

# E

lem

ents

Common modeling errors

Results analysis

Explicit feedback on language variant complexity

Language Variants

# op

inio

ns

Easy Med Hard

Rule of “thumb” on Language Variants

Variant 1Simpler, faster, less errors, limited power (no conditions)

Variant 2Strong thanks to looping, a lot of errors

Variant 3Good compromise. Limited by single local parameter

Variant 4Harder, slower, more errors. Multiple local parameters not appraciated

Rule of “thumb” on Single Elements

• Event Until

• Parallel

• Condition

• Cycle

• Global params

• Event Wait

• Local params single

• Local params multiple

Conclusions

• Simplification in mind• Definition of a formalized selection process of

language constructs and variants• Actual selection of a variant for our case study

Future Work• Modeling through multiple expertise levels

– From expert to the crowd

END OF THE STORY

NOT THEEND OF THE STORY

A company has born

www.fluxedo.com

Key: reduce complexity. FROM THIS…

TO THIS…

Integrating people…

Integrazione servizi online.. and Online Services

Take home message

• Being a modeler is hard• Modeling simplification seems really to lead to

extreme solutions, i.e. completely hide modeling

The challenge is not to show off modeling, is to hide it

An Empirical Study on Simplification ofBusiness Process Modeling Languages

[email protected] @marcobrambi@fluxedo_app

Thanks