Records Management Support in the Interoperability Framework for the

Portuguese Public Administration

Catarina Viegas

, Andr

e Vasconcelos

1,2

, Jos

e Borbinha

and Zaida Chora

INESC-ID, Instituto Superior T

ecnico, Avenida Rovisco Pais 1, Lisbon, Portugal

Administrative Modernization Agency, Rua Abranches Ferr

ao 10, 3, Lisbon, Portugal

Keywords:

Information Management, Public Administration, Interoperability, Records Management Metadata, Canonical

Data Model.

Abstract:

The Portuguese public administration has a core technological infrastructure for interoperability, which assures

reliable core transactions, but takes all information objects as equals, leaving any necessary specialization to

the applications. However, public administrations are highly regulated environments, which implies business

processes involving entities of that domain are subject to strong requirements for information management.

Records management in special is a speciﬁc concern, meaning metadata for that purpose must be produced

along the production of the regular business information objects. In that sense, when two or more entities of a

domain of this kind engage in transactions, it is helpful for all those involved if also metadata created for that

purpose can be shared, which requires it to be commonly understood. In Portugal, national guidelines have

been developed to support that goal, remaining now the challenge of their implementation. This is a classic

problem of interoperability in distributed information systems, which has particular challenges when scoped

in the domain of a large public administration, involving thousands of local systems. This paper describes the

results of a research project intended to provide a proof of concept for that for the case of the Portuguese public

administration, which resulted in a case of application of the Canonical Data Model method. The metadata

schema produced is assessed using the Bruce-Hillman metadata quality framework, which made possible to

conclude by its effectiveness, along with suggestions for future improvements.

1 INTRODUCTION

Organizations within the same public administration

exchange information frequently among them. This

information, which used to be mainly in the form of

physical paper-based documents, tend to be now busi-

ness objects when the transactions are supported by

digital information systems. However, these business

objects need to be kept as records in the sending and

recipient organizations.

Records are evidences of processes, making the

management of records within an organization ex-

tremely important. The Portuguese Public Adminis-

tration (PPA) is made of multiple organizations, each

one expected to manage its records according to its

speciﬁc regulations and requirements, while obeying

to a same general legal framework. For that pur-

pose, all entities are expected to have deﬁned spe-

ciﬁc records management systems (RMS), conceived

to capture, store and manage records (Barbedo and

Corujo, 2012).

The promotion of measures by the Portuguese

government to dematerialize business processes, led

to the development of an interoperability project

which could ensure the sharing of information

through the RMS of public organizations.

The existence of an interoperability infrastructure

for the integration of information systems for the

PPA, and common requirements for records manage-

ment previously deﬁned, motivates a solution for this

project, based on existing infrastructures.

This paper presents the results of the development

and validation of a data model for records metadata.

This document follows by presenting a description of

the current interoperability measures in the PPA, and

an overview of the techniques for ensuring systems

interoperability. These techniques are the basis for

the development of the solution, which is presented

in section 3. The results of the experiments performed

on the solution proposed are analyzed in section 4 and

the conclusions and future work are presented in sec-

tion 5.

Viegas, C., Vasconcelos, A., Borbinha, J. and Chora, Z.

Records Management Support in the Interoperability Framework for the Portuguese Public Administration.

DOI: 10.5220/0007673300840094

In Proceedings of the 21st International Conference on Enterprise Information Systems (ICEIS 2019) , pages 84-94

ISBN: 978-989-758-372-8

2 BACKGROUND

This section presents the most relevant research de-

veloped regarding the management of records, inter-

operability and their current role in the PPA. In this

section, are also presented the different integration ap-

proaches considered for this project.

2.1 Records Management and Records

Management Systems

According to the ISO 15489-1:2016 (ISO 15489-

1:2016, 2016), records are information that is cre-

ated, received and maintained as evidence of an orga-

nization’s business process and as an asset in pursuit

of legal obligation or in the transaction of business.

The same standard deﬁnes records management

as the ”ﬁeld of management responsible for the efﬁ-

cient and systematic control of the creation, receipt,

maintenance, use and disposition of records” (ISO

15489-1:2016, 2016), and thus accordingly it de-

ﬁnes Record Management Systems as information

systems that capture, manage and provide access to

records. According to the same standard, all records

should be deﬁned by metadata elements and every

system should have one or more metadata schemas

which state how to deﬁne a record.

Related work has been described in (Maguire,

2005), which depicts the implementation of a records

management system in the Estates Department of

the British Library, describing the decisions made

throughout its implementation process, including the

deﬁnition of an adequate metadata schema.

2.2 Interoperability in Public

Administrations

The Decision no. 922/2009/EC (European Parliament

and of the Council, 2009) deﬁnes interoperability as

the capability of two or more diverse public adminis-

tration (PA) organizations to interact by sharing in-

formation and knowledge through the exchange of

data between their information systems. The Euro-

pean Interoperability Framework for European Pub-

lic Services describes four levels of interoperability

(European Commission, 2010): legal, organizational,

semantic and technical. This research focus mainly

on achieving technical and semantic interoperability

across the PPA, promoting, consequently, the other

two levels.

Semantic interoperability is the capability of two

or more information systems to exchange informa-

tion, while guaranteeing that the information’s orig-

inal meaning is maintained after the exchange, in the

recipient system. The exchange of data across dif-

ferent information systems can face multiple barri-

ers, such as the lack of a commonly agreed metadata

schema, or divergences in interpretation of the data

exchanged (European Commission, 2017). There-

fore, the establishment of a common reference to be

used by every organization is crucial to achieve in-

teroperability in the domain of PA. The deﬁnition of

a metadata schema that is used by every organiza-

tion within the PA will facilitate the correct sharing

of metadata records every time two RMS engage in

a transaction. A metadata record is shared when it

is produced in a RMS and sent to another RMS, and

reused by the receiver to create a local record. MIP

is the current Portuguese metadata schema produced

to be applied by PA entities when managing their

records.

2.3 MIP - Metadata for Interoperability

DGLAB

, the entity that has the role of national

archive in Portugal, deﬁned MIP to support meta-

data interoperability for records management, with

the goal of deﬁning a common schema to be used by

public agencies to characterize their records.

MIP is a metadata schema, comprising 17 meta-

data elements, deﬁned to ensure semantic interoper-

ability within the PPA (Barbedo and Corujo, 2012).

By deﬁning a common schema to be applied by all

different PA entities to their records, the goal was

to ensure that the data exchanged was equally inter-

preted by every RMS of the PPA. This way, local

records, copies of the records in the original RMS,

could be automatically created in the recipient RMS,

aided by the data received. To identify the meta-

data elements important to be in the schema, re-

quirements from records management international

standards were considered. These standards state

the metadata elements each record should contain

to guarantee the record’s authenticity and reliability

(Barbedo and Corujo, 2012).

Even though the development of MIP was pro-

moted by the Portuguese government, it is not legally

mandatory for PPA organizations to use it in their

records. This generates a semantic problem across or-

ganizations. The use of the same data schema to char-

acterize records guarantees that records are rightfully

recognized, captured, stored and managed by any sys-

tem that supports the schema, achieving RMS inter-

”Meta-informac¸

ao para Interoperabilidade” in Por-

tuguese.

”Direc¸

ao Geral do Livro, Arquivos e Bibliotecas” in

Portuguese, which stands for General Directorate for Book,

Archives and Libraries.

Records Management Support in the Interoperability Framework for the Portuguese Public Administration

operability.

2.4 MEF/LC - Functions and Processes

in the Portuguese Public

Administration

MEF/LC is the result of the national project ASIA

(Lourenc¸o and Penteado, 2015) and consists in the

merge of MEF

and LC

. MEF is a classiﬁcation

scheme that constitutes a conceptual representation

of the functions performed by public sector organi-

zations, providing two levels of classiﬁcation. The

ﬁrst level represents the state functions and the sec-

ond the subfunctions in which the level 1 instances

can be divided (for example, ”Strategic planning and

management”

is a state function of level 1, with code

150, and ”Policy deﬁnition and evaluation”

is a sub-

function with code 150.10) (General Directorate for

Book, Archives and Libraries (DGLAB), 2013).

LC (General Directorate for Book, Archives and

Libraries (DGLAB), 2014) is a catalogue of the busi-

ness processes executed by the PPA. MEF/LC collects

the information provided by these two classiﬁcation

models, establishing a 4-level classiﬁcation scheme

to be used by the organizations as a referential in the

development of their own functional business classi-

ﬁcation schemes (Lourenc¸o et al., 2012). MEF/LC is

summarized by a table of codes that deﬁne the func-

tions, subfunctions and business processes executed

by public agencies.

The main problem with MEF/LC is the lack of

mandatory legislation able to establish the use of this

classiﬁcation model by public organizations, in simi-

larity with MIP. Since it is not of mandatory use, only

a small number of organizations of the PPA use this

classiﬁcation model in their records. As a result, or-

ganizations can choose to apply this or other classi-

ﬁcation model, generating a discrepancy in the way

records are classiﬁed.

Multiple services offered by the PPA require a

collaboration between different public entities. This

collaborative approach is often achieved through the

exchange of documents among organizations, reason

why interoperability has always been a concern in

”Avaliac¸

ao Suprainstitucional da Informac¸

ao Ar-

quiv

ıstica” in Portuguese.

”Macroestrutura Funcional” in Portuguese

”Lista Consolidada” in Portuguese.

”Planeamento e Gest

ao Estrat

egica” (in the original

(General Directorate for Book, Archives and Libraries

(DGLAB), 2013))

”Deﬁnic¸

ao e Avaliac¸

ao de Pol

ıticas” (in the origi-

nal (General Directorate for Book, Archives and Libraries

(DGLAB), 2013)).

the PPA. Interoperability measures such as MIP or

MEF/LC were developed considering this collabora-

tive feature of the PPA. However, a closer analysis of

these measures allowed to conclude that they may not

be enough for the scope of this research. This work

will provide the PPA with an interoperability solution

that considers the measures deﬁned, whilst being ca-

pable of mitigating the ﬂaws they may possess.

2.5 Integration of Information Systems

In this subsection we will introduce the fundamental

state of the art of architectures for integration of in-

formation systems, in relation to the the integration

platform currently used by the PPA.

2.5.1 SOA and ESB Architecture

Nowadays, where integration is concerned, busi-

nesses opt for approaches like a Service-Oriented

Architecture (SOA) and an Enterprise Service Bus

(ESB). SOA provides the capability of designing the

business as a collection of application, where each

is responsible for one task within a business context.

ESB are integration platforms that allow the coordina-

tion of the interaction between different applications

from different sources (Chappell, 2004), by routing

messages from one application to another.

Often used together, these approaches may en-

counter limitations, specially regarding data integra-

tion. Data integration is the process of combining

different data from various sources to generate a uni-

ﬁed view of all the data intended. Data integration

can become very complex when using a SOA-ESB

approach, specially in large SOA projects, since mul-

tiple systems exchange data with one another but can

have different data deﬁnitions. To mitigate this prob-

lem, ESB offers message transformation, the process

of converting the data format of a message to another,

through the deﬁnition of mappings that correlate the

different data schemas with each other, which can

be applied in Point-to-Point Integration pattern or a

Canonical Data Model approach.

2.5.2 Point-to-Point Integration

A Point-to-Point integration technique requires, for

each service, the manual creation of a message trans-

lator for every application it interoperates with, estab-

lishing a translation per interaction. Therefore, each

different data schema is translated as many times as

there are different data schemas within a SOA. Any

changes in any of the data schemas implies chang-

ing its translation in every system it communicates

with. Figure 1 (left) depicts the number of message

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

Figure 1: Point-to-Point Integration (left) and Canonical

Data Model (right) (Hohpe et al., 2004).

transformations that would be required in a process

with six different applications and six different data

schemas.

2.5.3 Canonical Data Model Integration

The Canonical Data Model (CDM) methodology con-

sists of developing a data model known by every sys-

tem within the same ESB process. Each system is

responsible for deﬁning their own message translator,

from the CDM to its own data format, to interoperate

with another application of the ESB. This way, every

system only needs to develop one message translator,

instead of developing a message translator per differ-

ent data format within the ESB. Figure 1 (right) illus-

trates how a CDM supports the integration of an ap-

plication within the ESB. This approach ensures that

applications are able to interoperate with one another

as long as they are able to translate the CDM into their

own data format.

2.5.4 Comparative Analysis

Regarding the number of translation steps, Point-to-

Point Integration approach requires only one transla-

tion step in an interaction between two different appli-

cations, whilst the Canonical Data Model Integration

approach requires a double translation (Hohpe et al.,

2004): from the source application’s data format to

the CDM and other from the CDM to the target appli-

cation’s data format. Considering that each transla-

tion adds latency to the message ﬂow inside the ESB

(Hohpe et al., 2004), the need to introduce an extra

translation step can decrease performance results in a

Canonical Data Model approach.

Point-to-Point integration also requires a differ-

ent message translator for every application, which

increases the complexity of this process every time

a new application is added, or every time there is

a need to change the translation of a data format.

The use of a CDM reduces the complexity of mes-

sage transformation, and still guarantees the ﬂexibil-

ity and heterogeneity that characterize a SOA (Dave

Hollander, 2011). With a CDM approach, a new level

of indirection is added among applications’ individ-

ual data formats, making easier future changes in the

data format (being only necessary to update that ap-

plication’s translator) and the integration of new ap-

plications (since there is only the need to implement a

translator for the application and the common model

and not a translator for every application).

Regarding scalability, the two approaches present

limitations when a change occurs. In a Point-to-Point

integration, if there is a change in any of the data

formats, all translations need change. In a CDM ap-

proach, if a change in the common data model is re-

quired, every application needs to update their data

transform. On the other hand, the CDM can reduce

the complexity of this process by guaranteeing that,

when developing the model, the process is done with

the maximum level of abstraction possible, consider-

ing every application’s data format, while still ensur-

ing a response to the business needs.

The use of a Point-to-Point integration is difﬁcult

in large SOA projects, since it requires a large number

of message translators. Considering the diversity and

size of the PPA, the deﬁnition of a CDM was the ap-

proach chosen. Even if sacriﬁcing the performance,

by adding the second translation step required by the

CDM, the reduction of the complexity of the process

pays off in the end.

2.5.5 iAP - Interoperability Framework for the

Portuguese Public Administration

The Administrative Modernization Agency (AMA),

is the Portuguese agency responsible for promoting

modernization within the PPA. In that scope, it de-

veloped iAP

, the interoperability framework for the

PPA. The main objectives for this development were

to 1) simplify the communication between organi-

zations and business partners by streamlining busi-

ness processes and developing services, and 2) facil-

itate and minimize the costs and effort of developing

new business processes (Administrative Moderniza-

tion Agency (AMA), 2011). Although the platform

has four main components, only the functionalities

provided by the Integration Platform, a platform de-

veloped as a state-wide SOA, that provides a catalog

of services published and consumed by entities of the

PPA, will be explored. The invocation of these ser-

vices is mediated through the use of an ESB, provided

by iAP.

”Interoperabilidade da Administrac¸

ao P

ublica” in Por-

tuguese.

Records Management Support in the Interoperability Framework for the Portuguese Public Administration

3 SHARING RECORDS

METADATA IN THE

PORTUGUESE PUBLIC

ADMINISTRATION

Interoperability can only be achieved through a mu-

tual agreement on all basis: technical, semantic, or-

ganizational and legal. The solution developed focus

on technical and semantic interoperability, in the de-

velopment of a consensual data model to be applied

as the service interface for the new service in iAP,

which allows the exchange of metadata. A service in-

terface deﬁnes which data the service needs, describ-

ing the message format to be used for data exchange

among systems. A service interface can be deﬁned by

a WSDL or an XSD, reason why the CDM developed

was implemented as the XSD of the iAP service.

3.1 Solution Overview

iAP offers a set of services, all of them deﬁned by

a speciﬁc CDM, ensuring that all the information re-

quired is provided by the organizations who invoke

these services. The solution developed is a CDM for

a new service in iAP, which will allow the sharing

metadata among systems.

The development of this CDM was based on MIP

(see subsection 2.3). Although MIP has issues, as

stated and shown next in subsection 4.1, the infor-

mation it provides must be preserved and included

in the CDM proposed. Using MIP’s element deﬁni-

tions, each element of the CDM has obligation and

repeatability attributes that state if the element must

be present and if it can appear more than once in

the SOAP message generated when the service is in-

voked.

This section will be divided into subsections, each

one representing a different type of change MIP el-

ements suffered when represented in the CDM pro-

posed.

3.1.1 Equivalent Element Deﬁnition

This subsection refers to the elements that maintained

the deﬁnitions proposed by MIP, when represented in

the CDM.

Elements Aggregation (Agregacao), Subject (As-

sunto), Coverage (Cobertura) and DocumentType’s

(TipoDocumental) structure and meaning remain

identical to MIP, the only difference being the way

a document type is represented. In the CDM, rather

than following MIP’s deﬁnition of the element, by al-

lowing the designation of any value, a numeric code

was assign to each document type considered by this

research.

3.1.2 Addition of New Subelements

Although MIP speciﬁes the elements necessary for

a correct description of the record, it was detected,

throughout this work, a need to add subelements to

already deﬁned MIP elements, to complete the infor-

mation provided.

Figure 2: Elements with new subelements.

As shown in Figure 2, Description (Descricao)

and Title (Titulo) are two MIP elements to which new

subelements were added. ScopeAndContent (Am-

bitoeConteudo) and FormalTitle (TituloFormal) were

added to these elements to complete the information

already provided.

3.1.3 Deprecated subelements

Considering the research context and the circum-

stances in which MIP was developed, some of the

subelements presented in it are no longer valid or nec-

essary for the CDM. The reasoning behind this de-

cision is either their insigniﬁcance for the research

(MIP subelement Support), or the redundant infor-

mation provided by them (remaining subelements re-

ferred in Figure 3).

Figure 3: Elements that lost subelements.

Represented in Figure 3

, elements RecordDates

(DatasRecurso), Format (Formato) and Accessibility

(Acessibilidade) have all lost one or more of their as-

sociated subelements represented in MIP.

3.1.4 Redeﬁnitions of Subelements

In the development phase of the CDM, it was de-

ﬁned that the information transmitted by MIP ele-

The English translation of the names of the MIP subele-

ments presented in the table are of the responsibility of the

authors. The original names, in Portuguese, can be con-

sulted in (Barbedo and Corujo, 2012).

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

ments would be equally transmitted by CDM ele-

ments. However, changes to the elements were re-

quired to ensure that the original meaning of this in-

formation was maintained when captured by another

RMS.

The information provided by these elements is the

same, but the way it is structured is different, in or-

der to respond to the needs of automation of metadata

capture and record creation processes, promoted by

the proposed solution.

For identifying organizations, elements Identiﬁer

(Identiﬁcador), Producer (Produtor) and Receiver

(Destinatario) all apply the same structure to their

subelements. XOrganizationType (TipoOrganismoX)

and XOrganizationID (IDOrganismoX) are two types

of subelements that were introduced with the goal of

providing a normalization to the identiﬁcation of or-

ganizations. X represents the role of the organization.

The way organizations are identiﬁed in the CDM

changed when compared to MIP. The proposed CDM

considers the need for an automation of the pro-

cesses of capturing metadata and registering new lo-

cal records in the RMS. This automation can only be

guaranteed if the data received is normalized. Thus, it

was deﬁned that organizations would be identiﬁed us-

ing only one common method, a SIOE

code. SIOE

assigns to each organization an unique code. The type

of organization is stated in element XOrganization-

Type, with values raging from 1 to 3 (1 is a SIOE reg-

istered organization, 2 is a non-SIOE organization but

iAP-user and 3 is for the remainder that do not fall

in none of the other categories. The unique number

that characterizes the organization is hold by element

XOrganizationID.

3.1.5 New Metadata Elements

When developing the CDM, new metadata elements

were also introduced. Figure 4 displays all of the ele-

ments that are not represented in MIP and were added

to the CDM. The obligation and repeatability of the

elements is represented in the table and both features

will affect the way they are implemented in the XSD

of the service, as shown in subsection 3.2.

Figure 4: New elements introduced in the CDM.

The element Priority (Prioridade) was added to

”Sistema de Informac¸

ao da Organizac¸

ao do Estado” in

Portuguese.

provide the organization with the information regard-

ing the handling priority of the data received. Known-

Referece (VossaReferencia) is used only when the

metadata is sent as a response to a request. An in-

formation object B is considered a response when

its contents answers the contents of information ob-

ject A. This element is exclusively meant to indicate

which related object triggered the response object be-

ing sent. By using this element, the recipient RMS

would recognize B as being a response to A, inde-

pendently of how many objects were related to object

B, identiﬁed in element Relationship, a MIP element

who transitioned to CDM, whose functionality is stat-

ing which records are related and the type of relation-

ship they established with one another.

SpeciﬁcMetadata (MetadadoEspeciﬁco), is a new

subelement which provides organizations with a way

of adding extra information in the metadata. Other-

ClassiﬁcationCode (OutroCodigoClassiﬁcacao) was

introduced to provide organizations that do not use

MEF/LC as a classiﬁcation model, the capability of

identifying the classiﬁcation model used. The addi-

tion of this element was a necessity considering the

CDM was designed to promote the use of MEF/LC

by the organizations of the PPA.

3.2 XSD Implementation

For implementation, the new data model was applied

to the XML Schema Deﬁnition (XSD) of the iAP ser-

vice.

To represent the repeatability and obligation of

the elements, the minOccurs and maxOccurs indica-

tors are used to speciﬁes the minimum and maximum

times an element has to appear in the schema. To rep-

resent a repeatable element (i.e. an element that can

appear more than once in the SOAP message gen-

erated), maxOccurs’ value must be unbounded. To

represent a mandatory element, indicator minOccurs

must take the value of 1 if the element is mandatory,

and the value of 0 if the element is optional. List-

ing 1 depicts an excerpt of the XSD ﬁle where element

OtherClassiﬁcationCode is characterized as being op-

tional (minOccurs = 0) and repeatable (maxOccurs

= unbounded), as shown in Figure 4.

In the XSD, elements are portrait as

complexTypes if they have associated subele-

ments, or simpleTypes if not. Subelements who

have subelements, are represented by complexTypes

as well, generating an hierarchical architecture that

allows systems to comprehend which elements

cannot exist without their parent elements.

Elements such as Relationship or Identiﬁer have

subelements whose range of accepted values is lim-

Records Management Support in the Interoperability Framework for the Portuguese Public Administration

ited. RelationshipType (TipoRelacao) is a subele-

ment of Relationship that represents the type of re-

lationship established between two records. All types

of relationships considered are represented by a nu-

meric code, ranging from 1 to 12. To represent this

limited set of values, the restriction attribute is

used. As shown in Listing 2, element Relationship-

Type can only accept integer values ranging from 1

to 12, implemented by the use of minInclusive and

maxInclusive indicators.

Listing 1: Excerpt of the service’s XSD ﬁle for element oc-

currence.

<xs:element name=”OtherClassiﬁcationCode”

type=”OtherClassiﬁcationCode”

minOccurs=”0” maxOccurs=”unbounded”/>

Listing 2: Excerpt of the service’s XSD ﬁle for value re-

striction.

<xs:element name=”RelationshipType”

minOccurs=”1”>

<xs:simpleType>

<xs:restriction base=”xs:int”>

<xs:minInclusive

value=”1”/>

<xs:maxInclusive

value=”12”/>

</xs:restriction>

</xs:simpleType>

</xs:element>

4 RESULTS AND DISCUSSION

To evaluate the proposed CDM, two methods were

employed: 1) a comparison between MIP and the pro-

posed CDM, to determine and analyze the importance

of the CDM and the improvements presented by the

model, regarding MIP; and 2) an assessment of the

qualities possessed by the CDM developed, accord-

ing to the Bruce-Hillman Framework (Bruce and Hill-

mann, 2004).

4.1 Comparing MIP and CDM

The comparison between MIP and the CDM devel-

oped consisted on the application of the two data

models as metadata schemas of iAP’s service, in the

same scenario. By applying MIP as the metadata

schema, two types of problems were identiﬁed: 1)

Lack of rigid norms of application (Problem A) and 2)

Structural problems (Problem B). The CDM presents

solutions for these faults, as it will be described in this

section.

Problem A is summarized by the lack of con-

trolled vocabularies and limits to the range of values

of the elements in MIP. Controlled vocabularies are a

set of accepted values that metadata elements can hold

(Online Computer Library Center, 2013). MIP docu-

mentation provides examples of the values elements

can possess. However, this is not enough to guaran-

tee interoperability, because it is not certain that ev-

ery organization will use and understand the values

equally. With MIP, every organization can apply any

value they deem ﬁt whilst, to ensure semantic inter-

operability among different systems, every organiza-

tion must apply the same set of values deﬁned. This

lack of common values also hinders the automation

process. The CDM proposed introduces a set of con-

trolled vocabularies to provide organizations with a

limit set of valid values for the elements proposed.

Problem B arises from the structural problems of

MIP. These structural problems were found in three

crucial MIP elements. MIP element ”Identiﬁcador de

recurso” (Record Identiﬁer) is responsible for iden-

tifying uniquely a record. The problem with this

element is the way it is structured in MIP, which

does not ensure the uniqueness of the identiﬁer of

a record within an interoperability process. MIP el-

ement ”Relac¸

ao” (Relationship) also has structural

problems in subelement ”Tipo de Relac¸

ao” (Rela-

tionship type), which contains multiple subelements,

one for each type of relationship the record may es-

tablished with another. This structure is not efﬁ-

cient since most of the subelements would be left

blank.MIP element ”C

odigo de classiﬁcac¸

ao”’s (Clas-

siﬁcation code) structure constitutes a problem, with

users not being able to identify the classiﬁcation

model used to classify the record and, consequently,

hindering the interpretation of the record’s class by

the receiver system.

For a better understanding of the origin of these

problems and how the CDM provides solutions for

them, an example will be presented.

As illustrated by Figure 5, suppose that Entity A

wants to inform Entity B. Upon receiving the doc-

ument ﬁle and metadata (represented in Figure 5 as

an unique entity for a simple understanding), a local

record is created in the RMS of Entity B, from the in-

formation received. After analyzing the received in-

formation, Entity B develops a response document,

registers the related information in its RMS and sends

the response to Entity A. Without the solution pro-

posed, entities would send the document via e-mail

or letter, which would lead to unnecessary costs and

time spent, and could potentially lead to an ineffec-

tive management of records, since there is no guaran-

tee that a new record would be locally created in the

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

Figure 5: Example of ﬂow of documents through iAP.

RMS of Entity A upon receiving the document.

As stated, the use of MIP AS-IS as the canon-

ical data model of the service would lead to multi-

ple faults. The way records are identiﬁed using MIP

would generate a problem if used to identify records

in the context of this work. MIP offers only one el-

ement to hold the value identiﬁer of the record. This

element in MIP does not have any special structure to

ensure that the identiﬁer of the record, which must be

unique, remains unique when shared with multiple or-

ganizations within the same interoperability process.

To solve this issue, CDM introduces a new struc-

ture for the identiﬁer element, with three subelements.

Subelements GeneratorOrganizationType and Gener-

atorOrganizationID provide information about the or-

ganization that identiﬁes the record, while Documen-

tID holds the identiﬁcation code provided by said or-

ganization. This way, even if two entities have dif-

ferent records with the same identiﬁer, as each orga-

nization is identiﬁed with an unique code, the set of

the three subelements maintains the record’s unique

identiﬁer.

To identify the classiﬁcation model used, MIP

uses one single element, which can hold any value the

organization deems appropriate, without restrictions,

since MEF/LC is not of mandatory use. However, the

key for success resides on the receiver organizations

identifying correctly the class of the record. With-

out knowing which classiﬁcation model is used, that

is not possible. Thus, when developing the CDM,

an extra element (OtherClassiﬁcationCode) was in-

troduced, to provide non MEF/LC compliant orga-

nizations of informing the receiver system of which

classiﬁcation model is being used. However, the use

of MEF/LC is promoted by the CDM, through the re-

structuring of element ClassiﬁcationCode.

The use of MIP in this context generates a problem

in the way organizations are identiﬁed in the meta-

data. In its documentation, MIP provides examples

of values for identifying organizations. The problem

is that these examples are considered as a suggestion

only, and not as a norm, leaving to organizations the

responsibility of using any method they deem ﬁt for

identifying organizations, not guaranteeing a seman-

tic agreement across all organizations. Thus, in the

CDM, it was deﬁned that every entity would be identi-

ﬁed using SIOE or, if not present in SIOE, other code

that is on a database controlled by AMA, as stated in

subsubsection 3.1.4.

To identify the relationships established with the

record, MIP provides element ”Tipo de Relacao” (Re-

lationship type), which contains 11 subelements, each

one responsible for identifying a different type of re-

lationship. As stated, this is not efﬁcient. The CDM

provides a new structure for identify the relationship

type, assigning to each type a numeric code, to pop-

ulate subelement RelationshipType. The CDM also

introduces element KnownReference, to identify the

record who triggered a response. As shown in Fig-

ure 5, Entity A sends to Entity B an information ob-

ject with identiﬁer Oﬁcio2222 through iAP. Entity B

receives this information and generates a local record

in its RMS. As a response to Oﬁcio2222, Entity B

produces a new information object and sends it to En-

tity A as a reply (Oﬁcio3333). To establish this re-

lationship between information objects in the meta-

data, Entity B indicates in element KnownReference

Oﬁcio2222 as the object to whom Oﬁcio3333 is a re-

sponse to.

The results of this comparison helped identify the

need to develop a new data model and the reason

why MIP was not chosen to be applied directly to

the service. MIP provides the elements required in a

Portuguese metadata schema to correctly identify the

records, but it does not provide a well-formed struc-

ture to be applied digitally. MIP was developed fo-

cusing more on how to identifying records within an

organization and less in how these records would be

interpreted if their metadata was shared with a differ-

ent organization.

4.2 Assessment of the CDM Qualities

The Bruce-Hillman Framework (BHF) is a technique

used to assess the qualities of metadata schemas

(Bruce and Hillmann, 2004), deﬁning seven qual-

ities: completeness, provenance, accuracy, confor-

mance to expectations, logical consistency and coher-

ence, timeliness and accessibility. Each quality is as-

sociated with questions whose answers provide a nar-

rative score, as depicted in Figure 6.

According to the BHF, Completeness is the capa-

bility of the metadata schema to describe the object

as completely as possible, considering the project’s

resources. To measure this quality, the BHF presents

two questions. As shown in Figure 6, the ﬁrst ques-

tion can be answered afﬁrmatively, considering that

Records Management Support in the Interoperability Framework for the Portuguese Public Administration

the CDM was developed under the inﬂuence of MIP.

As concluded in subsection 4.1, MIP was not de-

signed to be applied to records that are meant to be

exported to other systems, since it does not ensure

that the identity of the record is maintained when ex-

ported

. However, MIP was developed as a measure

for describing records metadata correctly. Even if the

way MIP is structured is not ideal and would not gen-

erate good results if applied in this context, it is a

fundamental reference for CDM. This way, we can

sustain CDM supports effectively the creation of the

local record, since it follows the requirements from

MIP. When developing the CDM, and in similarity

with MIP, there was a consciousness that not every

element is always required, since many just provide

extra information that helps to identify the record, but

is not necessary seeing as other elements are capable

of providing enough. However, every element that

is considered crucial to describe the record is present

and is mandatory, meaning that the information pro-

vided by those elements is always transmitted, with-

out exception. With this in consideration, the second

question proposed by the BHF, shown in Figure 6 is

positive, presenting two examples, in column ”Com-

pliance indicator”, of elements crucial to deﬁne the

record’s identity.

Provenance, as a quality, is deﬁned as the capabil-

ity the metadata has of providing information about

its origins and changes throughout time. To assess

that, three questions are proposed, as in Figure 6. The

ﬁrst question tries to understand if the element set pro-

vides information about the responsible for creating

the metadata. The CDM is capable of providing this

information, considering that the responsible for key-

ing the metadata of the record is the generator organi-

zation. The next two questions analyze if the metadata

provides information of how the metadata was created

and if the metadata has suffered any transformations

since its creation. The CDM does not provide any

information about these issues. The CDM only reg-

isters information regarding the version of the record

but not changes in its metadata.

Another quality promoted by BHF is Accuracy.

Bruce and Hillman state that a metadata schema

should be accurate in the way it describes the data ob-

ject, by providing ”correct and factual” (Bruce and

Hillmann, 2004) information. To assess this, the BHF

proposes three questions. As mentioned previously,

The export of a records from one system to another oc-

curs in very speciﬁc business scenarios, due for example a

legal obligation, for preservation of the original records due

to the decommission of an old system, etc. Anyway, even

if that is not a speciﬁc concern of this work, the results here

presented also can contribute for that to be more easily done

in the future over the iAP.

Figure 6: Bruce-Hillman framework applied to the CDM

(Bruce and Hillmann, 2004).

when developing the CDM, there was a concern in

ensuring that the information provided by MIP was

maintained by the new elements of the CDM. Con-

sidering this, the ﬁrst question, shown in Figure 6,

can be answered afﬁrmatively, seeing as MIP is an

accepted method and was used for the creation of the

CDM. The second question inquiries about what has

been done to ensure valid values and structure for el-

ements of the metadata schema. While developing

the CDM, a great deal of importance was given to the

structure of the elements, since this was the main fac-

tor why MIP was not apt to be used, as shown in sub-

section 4.1. For deﬁning the CDM, a set of profes-

sionals, from different areas within the community,

that would bring different ideas and perspectives to

the table, were selected to help evaluate which val-

ues were needed for each element and what structure

the elements should adopt to achieve the goals of the

research. Since the CDM has yet to be used by orga-

nizations within the PA, the third question cannot be

answered due to the lack of information regarding its

application.

A metadata schema is in Conformance to expec-

tations if it is able to respond to the users necessities,

by including elements that the community expects to

ﬁnd, while remaining realistic about what is and is

not important to be included. To evaluate this qual-

ity, BHF proposes three questions. The ﬁrst, shown

in Figure 6, can be answered afﬁrmatively. The deﬁ-

ICEIS 2019 - 21st International Conference on Enterprise Information Systems

nition of the CDM was driven by the need to guaran-

tee that a record would be described correctly whilst

ensuring that this information would not be lose its

meaning when received by another system. Consider-

ing that CDM elements provide the same information

as MIP elements, an accepted standard for deﬁning

records within the PPA, it can be said that the CDM

describes what it claims. The second question in-

quiries about the use of controlled vocabularies and if

they are aligned with the needs of users and records.

When developing the controlled vocabulary for ele-

ment DocumentType, a survey was performed to un-

derstand which types of documents were more fre-

quent within the organizations of the PPA. With this

information, it was possible to deﬁne a controlled vo-

cabulary with 38 types of documents to be used as

values for the element. This serves as an example of

how users’ needs were taken into account when es-

tablishing controlled vocabularies for each element.

The third question refers to the compromises made

throughout the implementation phase, and their regis-

tration. To elaborate the CDM, different opinions of

archivists and information systems professionals had

to be balanced, requiring multiple compromises, doc-

umented in meeting minutes and other internal docu-

mentation produced during the development phase of

the CDM.

Logical consistency and coherence are qualities

of a metadata schema with a consistent structure

throughout its deﬁnition and associated application

proﬁle. An application proﬁle is a set of metadata

elements with guidelines and policies associated that

indicate how the metadata elements are to be applied

to the objects. In this case, the CDM proposed is

throughly explained in its application proﬁle, expos-

ing to the users the accepted values and utilization

norms of every element. As shown in Figure 6, two

questions are associated to this quality. The ﬁrst,

questions if the data in elements is consistent through-

out. For this assessment, it is possible to evaluate

what the CDM offers, to ensure that the data will

be consistent throughout. This metadata schema is

consistent in the way it describes different elements

who have the same functionality in the schema, even

if used in different contexts. Elements of type XOr-

ganizationType and XOrganizationID have the same

functionality of identifying organizations, present the

same structure and accept the same type of values, but

can be applied to different types of organizations and

their different roles for the record. Even if there is

not, yet, a set of data that can be assessed for its con-

sistency, the CDM has provided all the tools to ensure

this consistency. To answer the second question pro-

posed, shown in Figure 6, taking into consideration

the lack of application of the CDM, it is only possi-

ble to compare the architecture of the CDM with the

architecture of MIP, and how consistent the data pro-

vided by both data models would be. As stated in sub-

section 4.1, MIP does not provide rigid rules for the

data as the CDM does. Although the MIP provides

multiple examples of values to assign to each element,

nothing prevents users of applying different rules in

the same metadata object. Consequently, this does not

ensure that the data will be consistent throughout.

According to the BHF, Accessibility is the capa-

bility a metadata set has to be viewed and compre-

hended. This quality is assessed by three questions.

The ﬁrst, questions the appropriateness of the ele-

ment set for the community. As stated previously,

MIP was used as the basis for developing the CDM.

Considering this, the CDM proposed is appropriate

for the community, taking into consideration the con-

cerns from multiple perspectives and providing a so-

lution for them. The second question inquiries about

the costs of maintenance of the element set. Consid-

ering the use of XML and the results from simula-

tive calculations, it was concluded that maintenance

is affordable and even preferable to the costs of cor-

respondence, nowadays, in the PA. The last question

audits the ease of adding further elements to the set.

The CDM provides element SpeciﬁcMetadata, which

allows the temporary addition of data to the metadata.

A more permanent addition to the CDM is also easy

to achieve due to the ﬂexibility of XML, the language

used for implementing the CDM.

Considering the data available, it was not possible

to evaluate the Timeliness, another quality promoted

by the BHF, of the CDM since it refers to metadata

and controlled vocabularies updates, which have yet

to be tested.

5 CONCLUSIONS AND FUTURE

WORK

The development of the CDM is the ﬁrst step to

ensure interoperability among records management

systems within the PPA. This is a decisive step for

achieving interoperability since it guarantees seman-

tic interoperability among organizations, assuring that

they ”speak the same language” and that the metadata

exchanged is descriptive enough, and well structured,

to enable its rightful interpretation and the creation

of a new record in a new RMS, based on the infor-

mation provided. By using MIP as the basis for the

elaboration of this data model, all the necessary ele-

ments for a correct characterization of the records are

represented. MIP’s vocabulary and structural prob-

Records Management Support in the Interoperability Framework for the Portuguese Public Administration

lems are addressed in the CDM proposed, which can

be considered an improved version of MIP, ready

to be applied to information systems with records

management capabilities. The results show that the

CDM promotes interoperability through iAP, the Por-

tuguese interoperability platform, by ensuring that the

information exchanged maintains its original mean-

ing. Results also show that the CDM produced holds

5 qualities, named Completeness, Accuracy, Confor-

mance to expectations, Logical consistency and co-

herence and Accessibility.

Even though the element set produced relates only

with the PPA, this proposal is useful to understand

the steps in an information interoperability project,

with special attention on the process of establishing

a canonical data model to achieve semantic interoper-

ability in a SOA-ESB environment.

Considering that the CDM developed stands for

the initial steps of an ambitious project of interoper-

ating all of the different records management systems

of the PPA, there are still several steps to be taken to

achieve this goal. For the future, it is important that

tests are executed, specially an evaluation regarding

the level of automation that the system can acquire in

the capture of the document and metadata exchanged

and storage of new records by the systems, due to the

application of the CDM. The CDM may need future

improvements according to the feedback provided by

the organizations when in use. For example, if the

service is highly used to expedite invoices, the addi-

tion of element Price (Prec¸o) may be considered, in-

stead of using the SpeciﬁcMetadata element, to refer

the price associated with the invoice, every time.

ACKNOWLEDGEMENTS

This work was supported by national funds through

Fundac¸

ao para a Ci

encia e a Tecnologia (FCT) with

reference UID/CEC/50021/2019 and by the European

Commission program H2020 under the grant agree-

ment 822404 (project QualiChain). We would also

like to show our gratitude for the professionals at

AMA and DGLAB who contributed to this research.

REFERENCES

Administrative Modernization Agency (AMA) (2011). In-

teroperabilidade na Administrac¸

ao P

ublica, Proced-

imentos para Ades

a iAP [Interoperability in the

Public Administration, iAP Adhesion Guidelines],

Version 3.0. in Portuguese.

Barbedo, F. and Corujo, L. (2012). MIP: Metainformac¸

para Interoperabilidade [MIP: Metadata for Interop-

erability], Version 1.0c. General Directorate for Book,

Archives and Libraries (DGLAB). in Portuguese.

Bruce, T. R. and Hillmann, D. I. (2004). The continuum

of metadata quality: deﬁning, expressing, exploiting.

ALA editions.

Chappell, D. (2004). Enterprise Service Bus. O’Reilly Se-

ries. O’Reilly Media, Incorporated.

Dave Hollander (2011). Common Models in SOA.

https://statswiki.unece.org/download/attachments/

65372409/common model in soa wp.pdf (Accessed:

15-06-2018).

European Commission (2010). COM(2010) 744 ﬁnal, An-

nex 2 to the Communication from the Commission to

the European Parliament, the Council, the European

Economic and Social Committee and the Committee

of Regions: ”Toward interoperability for European

public services”, European Interoperability Frame-

work (EIF) for European public services.

European Commission (2017). Improving semantic in-

teroperability in European eGovernment systems.

https://ec.europa.eu/isa2/actions/improving-semantic-

interoperability-european -egovernment-systems en

(Accessed: 19-12-2017).

European Parliament and of the Council (2009). Decision

No. 922/2009/EC of the European Parliament and of

the Council on interoperability solutions for European

public administrations (ISA).

General Directorate for Book, Archives and Libraries

(DGLAB) (2013). Macroestrutura Funcional (MEF)

[Functional Macrostructure (MEF)], Version 2.0. in

Portuguese.

General Directorate for Book, Archives and Libraries

(DGLAB) (2014). Lista Consolidada: 3

s n

ıveis

em planos de classiﬁcac¸

ao conformes

a MEF [”Lista

Consolidada”: 3rd levels in business classiﬁcation

schemes according to MEF]. in Portuguese.

Hohpe, G., Woolf, B., and Brown, K. (2004). Enterprise In-

tegration Patterns: Designing, Building, and Deploy-

ing Messaging Solutions. A Martin Fowler signature

book. Addison-Wesley.

ISO 15489-1:2016 (2016). Information and documentation

– Records management - Part 1: Concepts and princi-

ples.

Lourenc¸o, A. and Penteado, P. (2015). A caminho da

ASIA – Avaliac¸

ao Suprainstitucional da Informac¸

Arquiv

ıstica [On the way to ASIA - Suprainstitutional

Evaluation of Archivist Information]. In BAD Na-

tional Congress, number 12. in Portuguese.

Lourenc¸o, A., Penteado, P., and Henriques, C. (2012). O

desaﬁo da interoperabilidade na gest

ao dos arquivos

da Administrac¸

ao: propostas do

org

ao de coordenac¸

nacional de arquivos [The challenge of interoperabil-

ity in Administration’s archive management: Propos-

als from the national coordination archives associa-

tion]. In BAD National Congress, number 11. in Por-

tuguese.

Maguire, R. (2005). Lessons learned from imple-

menting an electronic records management sys-

tem. Records Management Journal, 15(3):150–157.

https://doi.org/10.1108/09565690510632337.

Online Computer Library Center (2013). Using a Con-

trolled Vocabulary. https://www.oclc.org/content/

dam/training/CONTENTdm/pdf/Tutorials/Metadata/

Controlled Vocabulary.pdf (Accessed: 07-08-2018).

ICEIS 2019 - 21st International Conference on Enterprise Information Systems