Automatic Business Process Model Assembly on the

Basis of Subject-Oriented Semantic Process Mark-up

Alexander Gromoff

, Nikolay Kazantsev

, Pavel Shapkin

and Leonid Shumsky

National Research University Higher School of Economics, BPM Department, Moscow, Russia

National Research Nuclear University MEPhI (Moscow Engineering Physics Institute),

Cybernetics and Information Security Department, Moscow, Russia

Keywords: Enterprise Architecture, Semantic Mark-up, Real-Time Business Architecture, Formal Process Specifica-

tion.

Abstract: Business processes of large companies often lack the flexibility of their smaller competitors. This fact re-

sults in slower reaction to the changes in business environment which usually require to launch complicated

IT projects. Nevertheless many tasks faced in these situations by IT departments have standard solutions

which could be expressed in form of semi-formalized best practices and recommendations. Authors propose

a business process model enriched by the semantic mark-up that enables to represent these recommenda-

tions in form of fully formalized high-level process descriptions. Using the proposed technical approaches

concrete business process models can be generated from these descriptions. Authors discuss formal methods

of searching the relevant high-level descriptions as well as rating and evaluation of the generated concrete

models based on different criteria.

1 INTRODUCTION

Nowadays there is recommended process configura-

tion that can execute almost every goal of a compa-

ny target. For this process in turn there are several

possible realisations by means of different programs.

The key enterprise architecture metricise its ability

to connect new processes and IT-means for their

support. The moving towards this direction is done

from several aspects. Firstly, the approaches to im-

provement of the process goal setting and strategic

planning are developed. The goal, which is formu-

lated correctly should not only satisfy the future

owner by its form but also be a process-setting mod-

el of actions. For ensuring such level of goal decom-

position different techniques of semantic modelling

and semantic analysis of received constructions are

used. Secondly, there is a tendency of putting stand-

ard process models and IT means for their realiza-

tion into cloud and this tendency is currently

empowered. In general this is the development of

process modelling and process realisation automa-

tion. The usage of cloud technologies allows com-

posing difficult processes from standard components

and reference or etalon solutions for developing pro-

cesses realisation.

Both described approaches seriously increase the

agility of enterprise architecture and approximate it

to the real-time architecture, when any change of

strategic goal immediately is reflected in change of

settings or/and content of software used. The aim of

this work is to make next step toward conjunction of

both described approaches.

The case is that, when semantic modelling of

goal formalisation is widely used then in the sphere

of process modelling the situation is different. Sug-

gested elementary canonical processes are modelled

sufficiently informal and heterogeneously and the

semantic modelling is used at best only for descrip-

tion of inputs, outputs of the processes and for no-

tion of its pre and post conditions. This brings us to

the fact that from one point of view the task of

searching the solution becomes difficult and compo-

site process construction considerably lost in visuali-

sation and in understanding, as a result – in

acceptance. From another point of view, the task of

connection of the constructed process model with

technological wares also becomes difficult and une-

ven. The outcome of these disadvantages is that

there is a big lag between the model and strategic

goal changes and this changes implementation. This

lag appears as a result of the need to update the pro-

158

Gromoff A., Kazantsev N., Shapkin P. and Shumsky L..

Automatic Business Process Model Assembly on the Basis of Subject-Oriented Semantic Process Mark-up.

DOI: 10.5220/0005120001580164

In Proceedings of the 11th International Conference on e-Business (ICE-B-2014), pages 158-164

ISBN: 978-989-758-043-7

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

cess model (or develop new model) and then again

«link» the technological wares to this new model.

Currently situation in relatively big enterprises is

close to dramatic misunderstanding what to do and

where to go. Actually in 10 observed entities with a

staff from 5K to 40K of personnel the lag of imple-

mentation of the required changes in IT or work-

flows is about 6-9 months. That means when

actually it’s a turn of change one can guaranty no

planed change is required anymore since situation

has changed again and new changes are necessary.

Considered approach is in application of process

description formal model, which should allow inte-

gration of strategic goal’s semantic model and pro-

cess technological implementation model. For

description of process formal model it is suggested

to use the variant of π-numeration. This model al-

lows describing the technological aspects of realiza-

tion thus more time will be spend to the algorithm

description of connection the model with strategic

goal semantic model.

2 THE DESCRIPTION OF

REAL-TIME BUSINESS

ARCHITECTURE AND

COMPREHENSIVE CASE

STUDY

This part is devoted to the description of target state

of projected system and to the description of the case

study where described approach can be implement-

ed.

Breath of an architecture new wave smells as an

interdisciplinary combination of the approaches and

getting out of concept layer: influenced by IT infor-

mation flows are growing rapidly eroding the barri-

ers between the departments and even companies.

Possibility of generation ad-hoc real-time business

processes on the global cloud-based self-generated

business service basis, however, is not yet provided.

The feedback loop makes business processes as

visible for corrections as locally efficient. The archi-

tecture itself takes the responsibility for the global

efficiency and strategy goals achievement. Main

obstacle for such business development or reengi-

neering on a platform of basic or referential patterns

is a ‘human factor’, which is a key issue in resource

usage for such processes.

SOA, it seems, had an ability to overcome

threshold of human factor nihilism. Nowadays when

“SOA” term is mentioned, professionals interpret it

as IT architecture with a fixed set of integrated ser-

vices. Since “fixed” is a key issue here due to stand-

ardization and cost saving strategies, the question of

responding to business agility trend is one of the

most vital for SOA today. The concept of real-time

business architecture is in charge of rapid changes of

business requirements (Gromoff et al., 2012). The

list of authors suggested an alternative view on busi-

ness modelling using free services in order to meet

variable business demands by means of creating the

prototype of app store - is-store in network where

services of various vendors are collected and simul-

taneous feedback from the users that attached their

personal experience (rating) is received.

The state-of-the-art for today’s technological fa-

cilities of the Big Data centres, which are grown

faster than mushrooms in rainbow forest, has started

unique process of dictating new mental paradigm for

traditional business mentalities. NOW we have to

understand that not only referential models of busi-

ness processes, or even executable blocks of busi-

ness units can be simply bought from Clouds traders,

but much more, something what always considered

as a main assets of any business — expertise and

intellectual capital. This futuristic reality of auto-

mated business engineering could be considered as

an approach to the newer business vision, which is

dictated by modern technological abilities and

tendencies.

From this respect it’s easy to see that in such

business architecture only strategic targeting and

monitoring is a responsibility of the true human ex-

ecutive level while the rest is compiled automatical-

ly from patterns and basics, best practices, and

concepts ‘as to be best’. In a final phase of business

orchestration intellectual resources are selected from

Clouds and after legal formalities are switched into

action of the ready-to-go processes.

Moreover, the scheme of functioning of suggest-

ed service is the following:

1. The user formulates the goal in the most gen-

eral form in the natural language.

2. The formal goal statement is generated on the

base of domain ontology.

3. As a result of reflection on formalized goal, the

set of formalized solutions with weights by dif-

ferent criteria is build.

In the fact, each solution is a business process

model. At the same time the weights allow to make

the choice of concrete solution on the base of one or

another preference: the speed of goal reaching, cost

of solution, the quality of result and so on. The

search of solutions as well as the weights calculation

is done on the base of semantic mark-up, which is

included in the process description. In its turn the

AutomaticBusinessProcessModelAssemblyontheBasisofSubject-OrientedSemanticProcessMark-up

159

search of process description is done on the base of

reference models, which are created by the experts

or imported from external sources.

3 GOAL FORMALISATION

The semantic mark-up is needed in order to “query”

the set of executable canonical process models using

the formalised goal statement.

The formalised goal conjuncts the structure of

entities of setting task and the set of strategies – the

definition of process on the level of domain ontolo-

gy. It is based on the domain conceptual model

(Baader et. al. 2007), (Roslovtsev et. al. 2013) and

describes the entities executable by the process: the

sequence of the steps that should be done under

these entities; the description of input data, expected

and intermediate results. One of the main semantic

characteristics of goal is the description of the re-

quired resources and correlation of these resources

with the resources of organization. Such, the strategy

prescribes that for setting goal execution the em-

ployees are needed and suggests the description of

human resources of concrete organization.

In the formal strategy description there should be

highlighted the following elements:

 Entities and their types;

 Actions;

 Parameters of actions and their types;

 Preconditions;

 Post conditions;

 Effects.

Entities are objects, which interact with one an-

other in the stated task. An object can be e.g. the

organisation itself – in case of goal formulation – or

different data objects etc. The types of the entities

are the categories from the domain ontology.

Actions are different methods of interaction be-

tween entities. Each action is characterised by

source (subject) and target (object). Also, the action

can be connected with some set of parameters, pre-

conditions, post-conditions and the effects.

The set of action parameter values is the infor-

mation connected with the action besides the source

and target. When the source and target of action can

be only some entities, the action parameters values

can be also some numeric values, text and so on. In

many cases the action parameters values cannot be

determined prior the process execution, for example,

during the description of actions of data transferring

the data transferred should be entered into the sys-

tem by the user. In such cases the types of parame-

ters should be indicated, setting the limitations on

the allowed set of values without specifying the val-

ues themselves.

For addition the action description can include

pre-conditions, post-conditions and effects.

4 EXTENSION OF Π-CALCULUS

FOR INCLUDING THE

SEMANTIC MARK-UP

This part is devoted to the extension of the grammar

and rules of abstract machine of π-calculus. The

needed extension should solve two basic tasks. First-

ly, the process model with linked semantic mark-up

should allow differentiating the processes by their

characteristics for comparison and ranging the mod-

els. Secondly, the model process execution with se-

mantic mark-up should allow picking out the redox

that answers the set semantic characteristics.

At first we will shortly describe the formal mod-

el being used. The alphabet of π-calculus consists of

the following components:

 The set of names of calculus , designated by

lowercase Latin letters;

 The set of processes, designated by uppercase

Latin letters;

 The typed calculus includes atomic types des-

ignated by lowercase Greek letters.

π-calculus terms are inductively builds by adding

special prefixes to existing processes or by combin-

ing existing processes. These operations are de-

scribed by the following grammar:

∶≔|̅.









.

|











|





!

For better expensiveness of model ability we will

use poliadic variant of calculus (Milner, 1992),

(Turner, 1996), which describe the transfer and re-

ceiving the list of names, but not the pressed name.

Let us examine the data designation more de-

tailed. The 0 symbol is used to designate the empty

process, or the process that does not execute any

activity, The ̅







,…,





prefix designates the trans-

fer of names 



,…,



by the link x, the 







,…,





prefix designates the data receiving. The prefix of

data receiving bounds the received names during the

process-continue (); the set of free names of the

process designates as 







, the set of bound

names is 







, data receiving is in the substitution

of all inputs of received name into the received val-

ue: if 







. – Is the process that receives data and

the process ̅ transferes data then the result of their

communication will be the process 



/



. The

communication (receive-transfer) is done when the

ICE-B2014-InternationalConferenceone-Business

160

one process transfers and the second process re-

ceives data by one the same name. Except for the

prefix of data receiving there is an opportunity for

bounding the name in the process with the help of

creating the local link







. This entry bounds the

name x in the process P – this name becomes the

locally determined in the process: no external pro-

cess can interact using this name. Thus instruction is

used for determination of internal, protected or tem-

porary channel of data exchange.

In the π-calculus there determined one manner of

process interaction writing – their parallel execution.

The parallel-executed processes can interact by cer-

tain rules. The replication of processes means ! the

process for which there always can be recieved the

new copy. The example of such process can be the

data transmition from the system when by the re-

quest the system can provide the data set which is

connected with the names 



,



by the channel : -

this can be written by the process !̅







,…,





.

The notion of this process without the replication

will mean the one-time transmition, which cannot be

repeated in the frames of the process – this is the

difference between message-request transmition and

the transmition of stored master data.

4.1 Semantic Process Mark-up

Semantic mark-up of the process is the set of param-

eters of the same ontology where the formalised goal

is described. The mark-up shows the characteristics

of business processes at each step of its execution.

The mark-up helps to define which actions and with

which costs are in concrete process variant.

The request to the set of canonical processes is in

that for each strategy from formalised goal the pro-

cess model (or its part) is chosen. This model is

closer to strategy and for each chosen model the

difference between existing and needed resources of

organisation is calculated in accordance with this

model.

Several general directions will do the semantic

mark-up of process. Firstly, we will consider how in

that model the pre and post conditions can be repre-

sented. Under precondition we will understand some

predicate which validity shows if the process can be

executed. Under post conditions we will understand

the predicate failure to execute which will be a fail-

ure and the process should be terminated. These

predicates will be modelled using the terms of λ-

calculus (Boudol, 1989) coded (Milner, 1992), (San-

giorgi, 2003) in the process. For coding the terms the

operator

|

|

will be used, which assosiate the term

λ calculus with correct process of π-calculus. This

operator extends the grammar of π-calculus process-

es by the following manner, where the Λ is the ran-

dom term of λ-calculus:

∷⋯|

|

|

.

There are rules, which code the terms of λ –

calculus.













.













|!



,



|



|



























|



|

|









|



|

|







.



,







The first rule associates the variable with process

of transition of its value using the stated channel.

The abstraction is coded as a process which creates

the local link and immediately transfers it through

the stated channel and in parallel triggers the process

replication, which after receiving the input data and

channel by the link returns the value of the term M.

Application is coded in a form of sequence – in the

beginning the first term is coded, after that the sec-

ond term id coded and the application of calculated

function to the argument is executed returning the

result for a, point of this term installing.

The preconditions are written in the following

manner. If the process P can start its execution only

if the predicate Λ



, dependant from the parameters





,…,



, received by the link l, is known, it is writ-

ten as



,











,…,











|







.0



This process starts it execution only if the pre-

condition has been executed. By analogy the post

conditions are written. The coded predicate receives

data from executed process but calculate them ac-

cording to the described logic, i.e. can operate be the

entities of conceptual model.

For domain entities description that are included

in the process model, the following grammar exten-

sion will be used. Let us pick out the set of available

names of π-calculus subset ⊆ ofdifferent

bytwovariables.Wewillcalltheset∪as

thesetoflabels.Thelabelcanbeeitherthename

ofπ-calculus or the special flag τ, represents the

absence of known label. The difference of labels

from the ordinary names of π-calculus is in the rela-

tion of labels to connection of variables. The charac-

teristic of the labels is that the label can not be

connected and can not be α transformed. Adding the

labels the grammar of term construction is changed

as the following:

∶≔







,…,





.,



,



…,





⊆

∶≔







,…,





.,∈,







,…,





⊆





∶≔







, ∈





∶≔| ∷!

We will designate this grammar as Γ



. If process

P is represented in this grammar let us write this as



⊢. As it is seen from the grammar definition,

labels can designate either transmitted data or

AutomaticBusinessProcessModelAssemblyontheBasisofSubject-OrientedSemanticProcessMark-up

161

names-channels using which the data transmition is

done. The connection of entities of domain to con-

crete transmited data brings the sense of calculation

of some fixed objects in the domain. This objects

can be corresponding structure divisions, regulative

documents and other objects of domain that do not

depend on concrete execution. The usage of label as

the channel of data transition/receiving will be used

for process execution assessment. For doing this

with each label a lot of entities of domain that repre-

sent the using resources, defined in the strategy, will

be connected. The execution of reduction step, in-

cluded such label will be estimated as the usage of

this resource, all such costs are saved during the

process execution. This makes it possible to estimate

the process execution in the context of all resources,

defined in the frames of strategy. All calculated

costs could be estimated in accordance with certain

rules and criteria, defined in the frames of strategy.

For further concrete process description let us

define the term of context of execution Δ for term

|Γ



⊢. Under this context we will understand the

vocabulary the key of which is label ∈, and the

value – object of conceptual model, connected with

this label. The sense of this context is in the connec-

tion of abstract names of π-calculus with using ob-

jects without the violation of the structure that is

defined by calculus and semantics. The term of π-

calculus that is having the empty context or does not

include the values of each label, we will call the

structure process model 



. The term the context of

which contain the values for all labels we will call

the conceptual process model 



,Δ ∈ 



⇒Γ



⊢ &∀ ∈ 







∩→ ∈Δ

,Δ ∈ 



⇒Γ



⊢&∃∈







∩→ 

∈Δ

The difference between these two types is in the

non-redex terms of the structure model, which can

be reduced in the conceptual model by the functions

execution that are connected with labels from the

context of process execution that are coded in the

same way as predicates in pre and post conditions.

In addition to writing in the context of execution

for label will be defined the term if the name type. In

the classical typed π-calculus for each name the type

is assigned that is an atomic type or channel type

(Pierce and Sangiorgi, 1993), (Barendregt et. al,

1977). We will extend this term for supporting the

semantic mark-up – will call the full type of the

name the pair



,



, where the first element is the

ordinary type of calculus and the second element is

the type, defined by the semantic object, connected

with this label. For the names that do not have the

connection with the semantic object, the second type

equals «null» type of using means of semantic mod-

elling. These changes will paired with using extend-

ed type system (Barendregt, 1991) for functions

defined by λ-calculus terms.

By its nature the process modelling using the π-

calculus does not operate with the definitions of ac-

tions (steps of the process) however they can be used

in the form of requirements in the strategy and using

the request to lots of model can be formulated. In the

strategy, the action will be represented as an indi-

vidual object, which is characterized, by the type of

the entity received as the input, the type of the result

and, optionally, the resources that are used for this

action execution.

For supporting the requests that operate the term

of action we will use the following agreement. Ac-

tion in the term of π-calculus is the term that is con-

ceptual process model reducing to the process,

consisting from prefixes of data transition, parallel

composition and empty process. The execution of

actions starts from receiving the data of label or

from executing some function where at least one of

the arguments is the label. The example of action

will be the process 



















,



〈





,

〉

where the following interpretation in frames of the

considered domain can be presented: the information

system 



for new employee x creates the unique

code a and transmits the employee data with its code

for further legislation 



,



. This action will be

characterised by the type of recieving object (new

employee) and the type of returning object (empty

type as the process terminates by the local data

transmition). For additional data identification there

can be identified the action according to the context

of process execution after the action reduction. The

set of actions that the process contains from is de-

fined by the process reduction and the selection of

all appropriate for action definition sub processes.

Summarising the receiving extensions we will

get:

1) The semantic mark-up of the process allows to

use in the processes pre and post conditions,

regulating the process execution depending on

the executing objects;

2) The context definition in the conceptual pro-

cess model allows to estimate the «price/value»

of process execution;

3) For the model there can be received the set of

executing actions in the terms of canonical

model, the received set can be compared with

set in the strategy by the coincidence of general

characteristics;

ICE-B2014-InternationalConferenceone-Business

162

4) The extended typification of process allows to

define the applicability of the process for the

required data.

Complete complex of actions, that are included

into semantic mark-up allows to organise the marked

up process model in such a manner that to the set of

these model requests, compiled on the base of goal,

can be directed and to receive the optimal model..

4.2 Case Study Formalization

Let us return to our motivating example of staff re-

cruitment task. On the formal level the specified

goal there is some concept “staff recruitment” action

with input parameter “profile” and the result of it is

the legalisation of new employee on the certain posi-

tion.

Suppose that the reference process models base

contains the descriptions for the following processes

of staff recruitment:

1) Internal recruitment (in filial of the company

etc.)

2) Staff recruitment through the staff agencies.

In the simplest form these processes can be for-

malised in the following manner. We will assume

that the internal requests for the employees trans-

ferred to the  channel and the requests for agen-

cies are transferred through the  channel.

Moreover, the employee should be legalised on the

new position for that the ERP system  is used. If we

assign the profile description as  and the data of

found employee is , we receive the following mod-

els:

1) 



|.̅,

2) 



|.̅,

It should be mentioned that ,  and  are the

labels. The type of the systems  and  corre-

spond to the action that receives as the inputs the

descriptions of profiles and returns the description of

found employee. The type of the system corre-

sponds to the action of legalization of employee on

the certain position. In such interpretation the types

of processes mentioned match conforms to the stated

goal formalization and thus will be included into

resulted set of solutions.

The conceptual model of described processes

contains characteristics of used systems ,  and

. The most interesting are those characteristics that

define the weight of the solution. For example, let us

examine such parameters as time consumption and

the level of hired employee. The time costs on the

personnel selection inside the company can be lower

than in the case of agency involvement, however,

the search of employee through the agency potential-

ly allows to find a highest-level employee. Of

course, the examined two borderline cases can have

different variations different from each other by dif-

ferent weights. These variations can also be included

in the reference base.

On the base of data from conceptual model for

found processes the weight is calculated by the de-

scribed parameters which gives the opportunity to

user to make argumented decision towards one or

another solution.

5 CONCLUSIONS

We have shown how the cloud real-time business

architecture can be implemented by means of formal

business process specification in -calculus enriched

by semantic mark-up. The semantic mark-up links

business process models to the domain conceptual

model; it helps to find the processes, which match

the specified goal and also to weight different solu-

tions on different semantic criteria.

The considered example describes most simple

cases. The incorporation of pre- and post-conditions

into the process description allows to automate the

construction of complex solutions – solutions, which

combine several processes. These solutions corre-

spond to complex processes that are compositions of

other (sub-) processes. In these cases pre- and post-

conditions could be used to determine the sequence

of sub-process execution. Proposed approach can

simplify implementation of process in semantic web

SOA environment (

Martin et. al. 2007).

REFERENCES

Milner, R., 1992. Functions as processes // Mathematical

structures in computer science. Т. 2. № 02. P. 119–

141.

Milner, R., 1993. The polyadic π-calculus: a tutorial. :

Springer.

Pierce, B., Sangiorgi, D., 1993. Typing and subtyping for

mobile processes // Logic in Computer Science.

LICS’93., Proceedings of Eighth Annual IEEE Sym-

posium on., P. 376–385.

Sangiorgi, D., Walker, D., 2003. The pi-calculus: a Theory

of Mobile Processes. : Cambridge university press.

Turner, D., 1996. The polymorphic pi-calculus: Theory

and implementation.

Barendregt, H.P., Dekkers, W., Statman, R., 1977. Typed

lambda calculus. P. 618.

Barendregt, H., 1991. Introduction to generalized type

systems. J. Funct. Program. 1, 125–154.

AutomaticBusinessProcessModelAssemblyontheBasisofSubject-OrientedSemanticProcessMark-up

163

Boudol, G.,1989 Towards a lambda-calculus for concur-

rent and communicating systems // TAPSOFT’89. P.

149–161.

Baader, F., et. al., 2007. The Description Logic Handbook:

Theory, Implementation, and Applications. : Cam-

bridge University Press. 622.

Vladimir, R., Leonid, S., Nikolay, K., et al., 2013. A syn-

thetic approach to building a canonical model of sub-

ject areas in the integration bus // In the proceedings of

ISKO-Maghreb, 2013 3rd International Symposium:

IEEE Xplore : Marrakech, Morocco - 2013 - C. 1 - 7;

DOI 10.1109/ISKO-Maghreb.2013.6728118

Gromoff, A., Kazantsev, N., Kozhevnikov, D., Ponfilenok,

M., Stavenko, Y., 2012. Newer Approach to Create

Flexible Business Architecture of Modern En-terprise./

Global Journal of Flexible Systems Management, Vol.

13, No. 4, p. 207-215

Martin, D., Burstein, M., McDermott, D., et. al, 2007.

Bringing Semantics to Web Services with OWL-S.

World Wide Web 10, 243–277. doi:10.1007/s11280-

007-0033-x

ICE-B2014-InternationalConferenceone-Business

164