The role of semantics in eGovernment service model verification and evolution

AAAI Spring Symposium 2006

Ljiljana Stojanovic, FZI

Andreas Abecker, FZI

Dimitris Apostolou, University of Pieraus

Gregoris Mentzas, ICCS,

Rudi Studer, AIFB

2

Politicians define the law

Programmers write the code

Experts decide how to implement the law Process1 Process2 Process3Start End

End-users use a portal

How to cope with the changes?

X

The goal of the OntoGov project is to improve back-office processes by taking into account the whole lifecycle!!!

suppliers

customers

back-office

Semantic Web Service

ID

Name

Time

Date

City

OutputsInputs

Get Birthday

Information

Web ServiceWeb Service Description

3

I1I2

O1O2O3

WS1I1I2

O1O2O3

WS2

I1I2

O1O2O3

WS

I1I2

O1O2O3

WS3

Monitor a document version

Analyse which activity to change

Plan how to achieve the consistency

Execute the required changes

Document X is changed

Activity Y has to be changedResource Z is not needed

Ontology Evolution generates additional changes and propagates

Effector (programmer modifies a code)

X

X

Sensor (add entry in a log)

XX

X

Change Management Framework

4

Modelling e-Government services

Meta Ontologies define the schema i.e. the language for modelling the e-Government services

Domain-oriented Ontologies model the concrete e-Government services and all data relevant for these services

Administration Ontologies enable better management of e-Government services

5

OntoGov model

LegalOntology

OrganisationalOntology

LifecycleOntology

DomainOntology

OntoGov ProcessOntology

OntoGovProfileOntology

Life-EventOntology

6

Meta-ontology cluster

Legal Ontology defines the structure of the legal documents, which includes paragraphs, sections, amendments, etc.

Organisational Ontology models an organisation by defining its organisational units, roles, persons, resources etc.

Domain Ontology contains domain specific knowledge LifeEvent Ontology models the categorisation of the e-

Government services Process Ontology describes the elements for modelling the

process flow Profile Ontology contains metadata about e-Government

services and includes all previously mentioned ontologies Lifecycle Ontology describes the information flow and the

decision making process in the public administration

7

Process Ontology

It is based on the OWL-S Process Ontology We distinguish between the services and the control

constructs Services can be either atomic or composite services We define a standard set of attributes such as name,

description, version, status etc. There are specific requirements concerning

retraceability, realisation, security, costs, etc. each service can be associated to the laws it is based upon each service can be associated to several software

components that implement it (i.e. dynamic binding) it is possible to assign security levels to each service information about cost and time restrictions can be also

specified Consistency conditions are formally defined

8

Life-Event ontology

It is used for the classification of the e-government services

It includes concepts such as residential affairs, residential permissions, identification certifications, naturalization citizenship, moving, education etc.

It has been developed based on the existing standards for modelling lifeevents For example, the Swiss Standard eCH-001 (Best Practice

Structure Process Inventory - http://www.ech.ch) aims: to give an overview over all relevant e-government services

in Switzerland and to provide a consistent and standardized classification of the

services. The inventory comprises 1.200 e-government services

that are all services initialized by a citizen or internal administration processes

9

Legal ontology

It models the structure of the legal documents, which includes paragraphs, sections, amendments We have analyzed the structure of legal documents in

Switzerland, Greece and Spain We concluded that the legal documents have very similar

structure independently of the country they are defined for Even though different countries use different terminology to organize

their legal documents, all of them use three levels of abstractions However, each country can extend (i.e. specialise or instantiate) it

as needed It is very important to document the laws and regulations the

process is based upon: not only for the whole process but also for specific activities

By associating legislation to these services, it is possible to trace and propagate the effects that a change in the legislation (or administrative regulations) produces on the models of the administrative services

10

Organisational ontology

It describes the roles and areas of responsibility and capabilities within an organisation with respect to the activities of a process model

It models the structure of an organisation, its resources, know-how, etc. For example, we distinguish two types of resources:

human resources who perform an activity equipment (i.e. hardware, software etc.) that is occupied

by the activity

11

Lifecycle ontology

It describes the decision-making process in the public administration

It bridges the gap between decision making and realisation by providing means: for describing these decisions and formally stating reasons that motivate the design

decisions

It provides answers on the following questions: “How have the process design (e.g. regarding atomic

activities) and flow (e.g. regarding control constructs) been realized?”

“Why has a design decision been taken?”

12

Domain ontology

It encodes concepts of the public administration domain such as the “terminology” used in the e-government domain For example, it defines the type and structure of

documents such as passport

Input and output of an activity are represented using entities defined in this ontology

13

“Announcement of moving” e-government service modelled using the OntoGov model

Lifecycle aspects

Design Decision 1:Eligibility handling

Reason I:Citizen must have Swiss domicile in order to perform automatic registration/deregistration

Related Instance(s):SR 101

SR 210 Art. 22A – 26A

Legal aspects

Organisational aspects

Domain InformationDomain Information

Domain InformationDomain Information

Domain InformationDomain Information

Domain InformationDomain Information

Domain InformationDomain Information

Domain InformationDomain Information

Domain InformationDomain Information

Domain aspects

14

Role of ontologies Ontologies are used to model (formally and explicitly) e-

government services OntoGov Profile Ontology:

Used for advertising and discoveing e-government services Based on OWL-S Profile Ontology Extends it with e-government specific metadata

LifeEvent Ontology is used for the classification of the e-government services

OntoGov Process Ontology: Gives a detailed description of a process flow Based on OWL-S Process Ontology Extends it with:

e-government specific metadata (e.g. Legal Ontology) in order to enable better and easier management of services

set of rules: for defining consistency of a service description for resolving inconsistencies

Set of ontologies (i.e. Legal, Organisational, Domain, Lifecycle and LifeEvent) used for annotation of a service description

Are e-government specific Model different part of reality Contain only the top-level entities

Each public organisation can extend it e.g. by specializing concepts or by creating instances

15

Ontologies are used to model aspects relevant for change management: Service Evolution ontology

It models what changes, why, when, by whom and how are performed in a service description, e.g.

Hierarchy of possible changes is developed based on the OntoGov model

They build the backbone of the change management system They correspond to the “conceptual” operation that someone wants to

apply without understanding the details (i.e. a set of ontology changes) that the management system has to perform

To enable reversibility as well as the synchronisation between different version of an ontology, this ontology includes:

the “cause-effects” dependency between changes a change may cause new changes in order to keep the service

model consistent the order in which the changes are requested

groups of changes of one request (requested or induced) are maintained in a linked list using the

Role of ontologies (cont.)

16

Change Management Framework

OntologyEvolution

ChangeDetection

MM

AAPP EE

Evolution Ontology

Process1 Process2 Process3Start End

Law

Code

AA

E-GovernmentPortal

MM

Serviceontologies

Usage OntologyLifecycle Ontology

Evolution Log

Usage Log

17

Change management

It has to enable the resolution of a given change in a systematic manner by ensuring the consistency of the whole ontology

InconsistencyDetection

ChangeGeneration

ChangeApplication

Change Resolution

Inconsistency Detection: It is responsible for checking of the consistency of an ontology with the respect to the ontology consistency definition. Its goal is to find ”parts” in the ontology that do not meet consistency conditions

Change Generation: It ensures the consistency of the ontology by generating additional changes that resolve detected inconsistencies

The approach requires: explicit specification of changes that can be applied the consistency definition

Changes have to preserve the consistency

18

Service consistency

Ontology consistency in general is defined as a set of conditions that must hold for every ontology

An e-Government service ontology is consistent: if it is ontology consistent and if it satisfies a set of consistency constraints defined for

the service model Consistency constraints have been defined in order to take

into account specificities of the Meta Ontologies Meta Ontologies represent the language for describing

services and therefore they define consistency of e-Government services

19

Inconsistency detection

Two approaches are possible: The procedural approach - semantics is given by

a procedural mechanism that is capable of providing answer to wide class of consistency problems It traverses the process model and checks every entity

within on its consistency Difficult to cover all options, since models may be very

complex The declarative approach is based on the sound

and complete set of consistency rules (provided with an inference mechanism) that formalises the model It applies reasoning based on a set of consistency rules

20

Incosistency detection

21

Changes

Ontology changes include changes such “AddAxiom” and “RemoveAxiom”

To make a service s1 a predecessor of a service s2, a domain expert needs to apply a list of ontology changes that connects s1 to s2

Domain experts require a method for expressing their needs in an exacter, easier and more declarative manner

For a domain expert it would be more useful to know that he can connect two services rather than to know how to do that

Intent of changes has to be expressed on a more coarse level

22

Semantics of changes is specified It guarantees that a required change is correctly

propagated and that no inconsistency is left in the system Each change is described in a form:

Precondition - a set of assertions that must be true to be able to apply a change

Postcondition - a set of assertions that must be true after applying a change and it describes the result of a change

Actions are additional changes that have to be generated

E.g. RemoveAtomicService X Precondition – AtomicService X has been defined Postcondition – AtomicService X doesn‘t exist anymore Actions:

Remove all input links of AtomicService X; Remove all output links of AtomicService X; Remove all metadata defined for AtomicService X that includes:

the attributes such as name, description, fist and last service; the relations to the legal, organisational and domain ontology; the pre- and post-conditionsX

Changes

23

Change propagation

Two types of change propagation are supported: From the associated ontologies to the service description From the included composite services to the service description

The following procedure is realized: The definition of a process model is extended with the version of

each referenced ontology Changes in the referenced ontologies are logged in their logs

(which results in the new version of these ontologies) Push-based synchronisaton: On the explicit request all changes

between two synchronisatons are discovered Their impact on a service model is determined For each „link“ the corresponding Remove change is

recommended For new entities the LifeCycle aspects are taken into account The domain expert may accept recommendations or not

24

OMS

25

State-of-the-art

26

OntoGov achievements

27

OntoGov SWS Activities

Ontology Management Publishing Discovery

28

Ontology Management

Within the OntoGov project, we have extended the standard approaches in two directions: Change Management – it is the timely adaptation of a

service description to the changes in business requirements, users’ needs, etc. as well as the consistent propagation of these changes to dependent artefacts The OntoGov approach enables agile response to frequent

and huge changes by ensuring the consistency preservation as well as the propagation of changes;

Lifecycle Management – even though knowledge is becoming recognised as one of the most important success factors of engaging policy makers, public administration managers, citizens and businesses in e-government, there is a lack of proven methods for the application of knowledge technologies The OntoGov lifecycle management bridges the gap

between decision making and realisation by providing means for describing these decisions and formally stating reasons that motivate the design decisions

29

Publishing

Publishing OntoGov service is based on the OntoGov Profile Ontology

Our approach focuses on imprecise or redundant annotation

It contains guidelines for building well-formed service announcement that are easier to be understood and cheaper to be modified

The “quality” of the annotation can be assessed through the existence of redundancy, inaccurate or incomplete information

30

Discovery

A number of proposals for automating the discovery of services are available They do not consider the fact that a user’s query is just an

approximation of his information need OntoGov service discovery has been realized by combining

three different types of conditions: Query-by-example– it is used for specifying conditions on the

service definitions, supplying the constraints on various fields; Reasoning using class hierarchy – the user can specify the type

of services. Subsumption reasoning is used to locate services that are more specific than specified;

Conceptual Query Refinement – the user defines keywords specifying the relevant terms that the service description must contain. The refinement system takes as the input the results of keyword-based search; whereas results are service descriptions. The system calculates refinements and generates a set of possible extensions of the original query

31

Architecture The novelty of the OntoGov approach lies in the formal verification of the service description as well as in the using of formal methods for achieving consistency when a problem is discovered This has been realized through two components:

Verificator Verification of the OWL-S process model is extended in several

dimensions First, we model the not only control-flow and data flow consistency

constraints. We allow to the public administrators to specify arbitrary domain-dependent consistency constraints. In this way we are able to cover all perspectives of the business models, i.e. control flow, data flow, operational issues (e.g. interactions between systems) and resources (e.g. humans, machines etc.)

Second, we do not consider only the process model but also the profile of a service

Finally, we have realized the verification of the e-government service descriptions using rule-based inference process

Recommender We propose formal approach for suggesting fixes that directly point to

the source of the inconsistencies The recommender is based on the so-called change-dependency graph,

which is a common technique for the maintenance of the knowledge-based systems

32

Service Ontology The most similar approach to the OntoGov approach is the OWL-S, since other approaches do not include the ontologies for modeling processes The OntoGov model consists of two major parts –

the OntoGov Profile ontology and the OntoGov Process ontology, which are developed based on the OWL-S ontologies

However, both of them are extended / adapted in order to take into account unique characteristics of the e-government services as well as some aspects needed for the better management of changes

33

Conclusion

Ontology-based change management system enables: automatic identification of inconsistencies in the description of

the E-Government services (log) analysis of a problem (lifecycle ontology) generation of recommendations for resolving problems

Advantages: Faster and better service design by all stakeholders involved

in the service lifecycle (e.g. managers, software developers) Better control and propagation of changes (e.g. control of

changes in law and propagation of changes to the software that delivers the service online)

More and better information about each step of the service delivery process, for all stakeholders involved in the service lifecycle

34

Thanks!Any questions?

Back up slides

36

User-defined Consistency

The user-defined consistency constraints are user‘s requirements that need to be expressed “outside” of the ontology language itself

Two types of the user-defined consistency conditions are identified:

generic conditions that are applicable across domains and represent best design practice or modeling quality criteria

modeling quality conditions: redundancy, misplaced properties, missing properties, etc.

domain dependent conditions that take into account the semantics of a particular formalism of the domain

All entities in the model must be connected Inputs & outputs must be defined in the domain ontology If input of a service is output of another service, then it has to

be subsumed by this output

37

OMS – Service Modeller & Service Registry

http://wim.fzi.de:8080/ontogov/ontogov.jnlp

38

Change generation

39

Change propagation

Extracting Deltas Reading Evolution Log

Analysis of changes For a new amendment a corresponding law can be found For a law all services that realize it can be found

Making Recommendation Lifecycle Ontology describes design decisions and their

relationship to affected parts of the service as well as to the requirements that motivate the decisions

It is a description of the service design process, which clarifies which design decisions were taken for which reasons, proves to be valuable for further development and maintenance

Legal

Legal‘

Meta

OntoGov

Supplier