MDE-Based Graphical Tool for Modeling Data Provenance According to

the W3C PROV Standard

Marcos Alves Vieira

1,2 a

and Sergio T. Carvalho

2 b

Instituto Federal Goiano - IF Goiano, Brazil

Universidade Federal de Goi

as - UFG, Brazil

Keywords:

Data Provenance, W3C PROV, Graphical Modeling, Metamodel, PROV-N, Modeling Tool, MDE.

Abstract:

The rise of the Internet of Things (IoT) and ubiquitous computing has led to a signiﬁcant increase in data vol-

umes, necessitating robust management. Data provenance is crucial for ensuring data reliability, integrity, and

quality, tracking the origins, transformations, and movements of data. The W3C PROV standard, with syn-

taxes like PROV-N, provides textual and graphical representations for expressing and storing data provenance.

However, despite its importance, there is a lack of user-friendly graphical tools for developers, particularly in

IoT and ubiquitous computing. This paper addresses this gap by introducing an innovative graphical tool that

enables the creation of user-friendly graphical data provenance models adhering to the W3C PROV standard.

The tool offers an intuitive interface for developers, simplifying the process of obtaining PROV-N code from

the generated provenance graph. We demonstrate the tool’s versatility across diverse domains, emphasizing its

role in bridging the gap in graphical provenance modeling. The paper outlines the Model-Driven Engineering

(MDE) methodology used in the tool development, and introduces its underlying Ecore metamodel aligned

with the PROV data model (PROV-DM). Evaluation results of the metamodel are presented, and potential ap-

plications of the tool are discussed, emphasizing its contribution to enhancing provenance-aware applications.

1 INTRODUCTION

The Internet of Things (IoT) and ubiquitous comput-

ing have revolutionized our digital interactions, gen-

erating vast data volumes requiring effective manage-

ment. Data provenance, tracking data origins, trans-

formations, and movements, is critical for ensuring

reliability, integrity, and quality in these domains (Hu

et al., 2020; Herschel et al., 2017).

The W3C PROV standard, encompassing the

PROV-DM data model, offers concrete syntaxes for

representing, serializing, and storing data prove-

nance (Moreau et al., 2013a). PROV-N provides tex-

tual representation, while a graphical layout follows

standard conventions in a directed graph.

When instrumenting software for data prove-

nance, developers must model captured object infor-

mation and determine the appropriate application ﬂow

time. For instance, in an IoT-based security system,

tracking conditions triggering emergency alarms from

each sensor may be relevant.

https://orcid.org/0000-0002-5498-5015

https://orcid.org/0000-0002-4191-5173

Data provenance can also be obtained through the

PROV-Template approach (Moreau et al., 2018), in-

volving mining relevant data directly from databases

or log ﬁles. This bypasses the need for explicit track-

ing or recording of provenance information during ac-

tivity execution.

In IoT and ubiquitous computing contexts, data

provenance’s signiﬁcance is paramount. Regardless

of the capture method, there is a need to model the

process for an accurate representation of data and

their interrelationships. However, to the best of the

author’s knowledge, there are no tools available for

developers to create desired provenance models for

software capture in a graphical manner. This gap

intensiﬁes due to data ubiquity and complexity in

IoT and ubiquitous computing scenarios. Develop-

ers typically adopt notations like PROV-N from the

W3C PROV model, yet a lack of graphical tools

means developers must convert static PROV graphs

into PROV-N code without interactivity. Modiﬁca-

tions in the PROV-N model mandate regenerating the

corresponding PROV graph and vice versa.

This paper introduces a novel graphical tool ad-

dressing this gap, facilitating user-friendly creation

Vieira, M. and Carvalho, S.

MDE-Based Graphical Tool for Modeling Data Provenance According to the W3C PROV Standard.

DOI: 10.5220/0012354700003645

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 12th International Conference on Model-Based Software and Systems Engineering (MODELSWARD 2024), pages 141-148

ISBN: 978-989-758-682-8; ISSN: 2184-4348

141

of data provenance models compliant with the W3C

PROV standard. The tool provides a direct, intu-

itive, and interactive graphical interface for develop-

ers in provenance modeling. Additionally, the tool

streamlines obtaining PROV-N code from the gener-

ated graphical model.

The remainder of this paper is structured as fol-

lows: Section 2 explores the theoretical and techno-

logical underpinnings employed in this study; Section

3 discuss related works and highlights the contribu-

tion of our proposal; Section 4 presents the founda-

tional Ecore metamodel representing the PROV data

model, followed by the W3C PROV-compliant graph-

ical provenance modeling tool built upon this meta-

model; Section 5 showcases the tool’s application in

modeling data provenance across diverse domains; ﬁ-

nally, Section 6 provides concluding remarks.

2 BACKGROUND

2.1 Data Provenance

Provenance refers to information detailing the pro-

duction process of a product, be it a data entity or

a physical object (Herschel et al., 2017). The W3C

Provenance Working Group

deﬁnes provenance as

“information about entities, activities, and people in-

volved in producing a piece of data or thing, which

can be used to form assessments about its quality, reli-

ability, or trustworthiness.” The term originates from

the Latin pr

oven

ıre, meaning “coming from.”

Data provenance is crucial in various application

areas (Glavic, 2021; P

erez et al., 2018): in open in-

formation systems, it aids in determining data ori-

gins and identifying responsible entities; in science

applications, it provides insights into research result

derivation, ensuring transparency and reproducibility;

in news contexts, it is vital for verifying blog and

news item origins, enhancing credibility and trustwor-

thiness; in legal domains, provenance inﬂuences li-

censing, attribution, and privacy information manage-

ment. In IoT, data provenance spans supply chains,

health monitoring, digital forensics, and intelligent

IoT services (Hu et al., 2020). Across these domains,

provenance ensures accountability, authenticity, and

informed decision-making (Herschel et al., 2017).

2.2 W3C PROV Family of Documents

The PROV Family of Documents, introduced by the

W3C Provenance Working Group, encompasses a

https://www.w3.org/2011/prov/

comprehensive framework that includes a data model,

serializations, and supplementary deﬁnitions. It es-

tablishes standardized representations and interoper-

able mechanisms for seamless sharing of provenance

data across platforms.

In W3C PROV, provenance is represented by three

central ﬁgures: Entity, Agent, and Activity, connected

by various relationships designed for assertions about

the past, such as wasGeneratedBy, used, wasAssoci-

atedWith, and wasAttributedTo. These relationships

are deﬁned in the Provenance Data Model (PROV-

DM) (Moreau et al., 2013a), the conceptual model

underlying the W3C PROV family of speciﬁcations.

The PROV standard also introduces PROV notation

(PROV-N) (Moreau et al., 2013b) for concise, human-

readable PROV instances and serializations, with the

aim of simplifying the translation of the PROV data

model into concrete syntax.

Entities, activities, and agents can be graphically

represented using W3C PROV conventions

, form-

ing a provenance graph. This graphical representa-

tion is further exempliﬁed in Figure 1, depicting the

process of writing a journalistic article with reliance

on a government-provided database. In this example,

journalist Bob, represented as an agent, was assigned

the entity employment-article-v1.html, indicating his

responsibility for creating the article. The activity

writeArticle demonstrates that the article was written

with associations wasGeneratedBy and used, signify-

ing the usage of oesm11st.zip as a base entity. The

graphical representation and PROV-N notation code

were developed under the proposed graphical data

provenance modeling tool, detailed in Section 3.

2.3 Model-Driven Engineering

Model-Driven Engineering (MDE) is a software de-

velopment approach that revolves around creating,

manipulating, and utilizing models as the central ar-

tifacts throughout the development lifecycle (Buc-

chiarone et al., 2020).

Models are typically constructed using Domain-

Speciﬁc Modeling Languages (DSMLs), which, in

turn, are deﬁned by a metamodel (L

opez-Fern

andez

et al., 2015). DSMLs provide specialized languages

tailored to speciﬁc domains, allowing for the creation

of models that capture the essential concepts and re-

lationships within that domain. In this sense, a meta-

model is “a model that deﬁnes the structure of a mod-

eling language” (Rodrigues da Silva, 2015) and serves

as a formal description of the language’s syntax, se-

mantics, and constraints (Bruel et al., 2018).

https://www.w3.org/2011/prov/wiki/Diagrams

MODELSWARD 2024 - 12th International Conference on Model-Based Software and Systems Engineering

142

nowpeople:Bob

is:writeArticle

now:employment-article-v1.html

govftp:oesm11st.zip

wasAttributedTo

wasDerivedFrom

wasGeneratedBy

used

(a) Graphical representation (b) PROV-N notation

Figure 1: Example of a provenance record represented in W3C PROV. Adapted from: (Moreau and Groth, 2013).

2.4 Eclipse Modeling Framework

The Eclipse Modeling Framework (EMF) (Steinberg

et al., 2008) is a robust modeling framework built on

top of the Eclipse Integrated Development Environ-

ment (IDE). EMF offers a range of features for creat-

ing, editing, and validating models and metamodels.

One of its key capabilities is generating code from

models, allowing developers to create Java implemen-

tations that correspond to the metamodel’s classes.

This way, each metaclass in the metamodel can be

mapped to a corresponding Java class, enabling the

instantiation of these classes to create models that ad-

here to the metamodel.

In EMF, metamodels are created using the Ecore

meta-metamodel, which serves as the central lan-

guage for deﬁning metamodels within EMF (Stein-

berg et al., 2008). Ecore itself is based on the MOF

(Meta Object Facility) meta-metamodel.

2.5 Eclipse Sirius

Eclipse Sirius (Madiot and Paganelli, 2015) is an

Eclipse project that empowers the creation of cus-

tomized graphical modeling workbenches. This al-

lows architects to develop domain-speciﬁc modelers

tailored to their unique business domains. Utilizing

the Eclipse Modeling stack, particularly the Eclipse

Modeling Framework (EMF), Sirius provides archi-

tects with a platform to declaratively specify their de-

sired modelers using a description model. This spec-

iﬁcation encompasses crucial aspects, including do-

main element representation, visual styling informa-

tion, and behavioral characteristics. Sirius offers a

comprehensive environment for creating these spec-

iﬁcations, resulting in description models that seam-

lessly function as Eclipse plug-ins. Notably, Sirius

supports dynamic and incremental development, en-

abling architects to make real-time modiﬁcations to

modeler deﬁnitions, ensuring continuous adaptability

and responsiveness to the evolving requirements and

challenges of dynamic business environments.

2.6 Eclipse Acceleo

Eclipse Acceleo

is an open-source, template-based

source code generation technology by the Eclipse

Foundation. It enables the creation of custom code

generators with comprehensive integration into the

Eclipse IDE, featuring syntax highlighting, real-time

error detection, quick ﬁxes, and dedicated views for

code generation design patterns and navigation within

the code generator. Acceleo supports incremental

generation, allowing manual modiﬁcations to gen-

erated code while preserving changes during subse-

quent regenerations by deﬁning protected areas for

safe modiﬁcations. The technology also incorporates

the Acceleo Query Language (AQL) for navigation

and querying of Ecore (meta)models, offering ad-

vanced features such as robust validation and support

for calling methods deﬁned through Java services.

3 RELATED WORK

ProvToolbox (Moreau, 2016) is a Java library devel-

oped by the W3C Provenance Working Group. It cre-

ates Java representations of PROV-DM and facilitates

conversion between Resource Description Frame-

work (RDF), PROV-XML, PROV-N, and PROV-

JSON. Offering command-line utilities, APIs, and

graphical interfaces, ProvToolbox supports diverse

provenance-related tasks, including parsing and se-

rializing documents, consistency checks, visualiza-

tions, and integration into existing software systems.

The toolbox aims to enhance the adoption of prove-

nance across domains, empowering users to effec-

tively manage and leverage provenance information

for tasks such as data analysis, reproducibility, ac-

countability, and transparency.

PROV Python (Huynh, 2020) is a Python pack-

age designed for efﬁcient handling of provenance in-

formation. Centered on data provenance and analyt-

ics, it allows users to create W3C PROV documents

https://eclipse.dev/acceleo/

MDE-Based Graphical Tool for Modeling Data Provenance According to the W3C PROV Standard

143

in Python, export them to formats such as PROV-

N and PROV-JSON, and visualize them graphically.

The package also facilitates interaction with ProvS-

tore (Huynh and Moreau, 2015), a cloud-based prove-

nance repository.

While the tools discussed ensure compliance with

the W3C PROV standard and support provenance

recording and conversion between different formats,

they lack direct graphical representation of prove-

nance and the ability to convert from a graphical for-

mat to textual formats like PROV-N. Currently, no

available tool allows for graphical modeling of prove-

nance. The tool presented in the forthcoming section

aims to address this gap by offering intuitive and in-

teractive graphical modeling of provenance under the

W3C PROV standard. This tool streamlines record-

ing and extraction provenance from software systems.

Once graphically modeled, it is possible to export the

provenance record directly in PROV-N notation.

4 GRAPHICAL PROVENANCE

MODELING TOOL

Provenance modeling, akin to modeling other soft-

ware aspects, enables the visualization of dependen-

cies and relationships within data manipulated by the

software. Graphical provenance modeling is espe-

cially beneﬁcial, providing system developers with

an interactive and visual approach to handling prove-

nance models. Following model design, develop-

ers can generate corresponding PROV-N codes and

instrument the software to record provenance based

on established models and speciﬁed granularity pref-

erences, including decisions about what provenance

data to store, where within the application, and at

what frequency.

The development of the graphical data provenance

modeling tool followed a systematic three-stage ap-

proach: (1) converting the PROV-DM data model

into an Ecore metamodel, named Ecore4PROV-DM;

(2) constructing a graphical modeling tool based on

the developed metamodel; and (3) creating model-to-

text (M2T) transformation templates to convert prove-

nance models built in the tool into PROV-N code in-

stances. The subsequent subsections provide detailed

insights into each of these stages.

4.1 Ecore4PROV-DM Metamodel

In the ﬁrst stage of the graphical data provenance

modeling tool development, we transformed the

PROV data model into a cohesive Ecore metamodel,

serving as the foundational framework for subsequent

modeling activities. This conversion closely adhered

to the descriptions of W3C PROV components as out-

lined in (Moreau et al., 2013a), ensuring alignment

with established standards and guidelines.

The resulting metamodel, named Ecore4PROV-

DM and depicted in Figure 2, was constructed in EMF

to represent the PROV data model. The metamodel’s

semantics are deﬁned by relationships between meta-

classes, with multiplicities represented by edges and

their numbering. The metaclasses in blue constitute

the metamodel’s core: Entity, Activity, and Agent.

PROV-DM’s creators categorized its model ele-

ments into six components, each serving speciﬁc pur-

poses (Moreau et al., 2013a). The Ecore4PROV-DM

metamodel follows this component-based approach:

1. Entities and Activities: deﬁne relationships be-

tween an Entity and an Activity. This component

is represented in Ecore4PROV-DM in yellow.

2. Derivations: addresses derivations of entities from

other entities and subtypes of derivation. Repre-

sented in Ecore4PROV-DM in red.

3. Agents, Responsibility, Inﬂuence: addresses

agents and the relationships that represent their re-

sponsibility and inﬂuence over an entity or activ-

ity. Represented in Ecore4PROV-DM in green.

4. Bundles: refers to a mechanism for supporting

the provenance of provenance. Represented in

Ecore4PROV-DM in purple.

5. Alternatives: allows expressing alternatives or

specializations of entities. Contains the Entity

metaclass and two binary associations, which are

self-relationships of the Entity metaclass: alterna-

teOf and specializationOf.

6. Collections: concerns the notion of collection,

which is an entity that has some members. The

members are, in turn, entities and therefore their

provenance can be expressed. This component is

represented in Ecore4PROV-DM in orange.

The Attributes and AttributeValue metaclasses, de-

picted in grey in Figure 2, are used to instanti-

ate the attribute-value pairs that most of the other

Ecore4PROV-DM metaclasses can have.

4.1.1 Ecore4PROV-DM Metamodel Evaluation

A team of metamodeling experts, following the

Metamodel Quality Requirements and Evaluation

(MQuaRE) framework, as proposed by (Kudo et al.,

2020), evaluated the completeness, correctness and

usability of Ecore4PROV-DM in comparison to the

PROV data model (PROV-DM). MQuaRE outlines 19

Metamodel Quality Requirements (MQRs) associated

MODELSWARD 2024 - 12th International Conference on Model-Based Software and Systems Engineering

144

with 23 quality measures, guiding metamodel evalua-

tions.

The evaluation involved presenting evaluators

with two foundational documents providing context

for the metamodel domain, including an introduction

to data provenance concepts, a brief outline of PROV-

DM, and comprehensive insights into the implemen-

tation of Ecore4PROV-DM within the Eclipse Mod-

eling Framework (EMF). After reviewing these doc-

uments and the Ecore4PROV-DM class diagram, the

experts completed a questionnaire covering personal

details, educational background, knowledge of data

provenance and metamodeling, and speciﬁc inquiries

about Ecore4PROV-DM’s completeness and accuracy

concerning its alignment with the PROV data model.

The evaluation questionnaire aimed to assess three

Metamodel Quality Requirements (MQRs) outlined

by the MQuaRE framework. Speciﬁcally, it focused

on MQR02, evaluating the metamodel’s completeness

by assessing its coverage of concepts speciﬁed in the

PROV-DM data model. Additionally, it addressed

MQR03, examining the correctness of the metamodel

to ensure accurate representation of PROV-DM con-

cepts. Finally, MQR07, a usability requirement, mea-

sured the proportion of metamodel concepts that are

easily identiﬁable to users. These MQRs were cho-

sen for their alignment with research objectives, fa-

cilitating the evaluation of the metamodel’s accuracy,

comprehensiveness, and user-friendliness in correctly

representing PROV-DM concepts.

Fifteen experts, all holding degrees in Computer

Science, participated in the evaluation process. Of

these, one had an undergraduate degree, while the rest

held master’s degrees, PhDs, or were pursuing a PhD.

Regarding prior knowledge, 73.3% were unfamil-

iar with data provenance, 20% lacked knowledge of

Model-Driven Engineering (MDE), and 26.7% were

not acquainted with the Eclipse Modeling Framework

(EMF). Regarding the support documents, 93.4% of

respondents found the document on the fundamen-

tals of data provenance and PROV-DM easily under-

standable, and 80% expressed clarity in comprehend-

ing the document explaining the implementation of

Ecore4PROV-DM in EMF. The remaining evaluators

afﬁrmed that both documents were sufﬁciently clear,

with none indicating difﬁculty in comprehension.

The evaluation results demonstrate unanimous

agreement among the participating experts regard-

ing the seamless alignment of Ecore4PROV-DM with

the PROV-DM speciﬁcation (MQR02 and MQR03).

However, one specialist expressed reservations con-

cerning the modeling of the “End” concept (wasEnd-

edBy) within Component 1, and another specialist

raised concerns about the implementation of Compo-

Entity

Name : EString

id : EString

Activity

Name : EString

id : EString

startTime : EString

endTime : EString

Agent

Name : EString

id : EString

WasGeneratedBy

id : EString

time : EString

Used

id : EString

time : EString

WasInvalidatedBy

id : EString

time : EString

WasInformedBy

id : EString

Attributes

WasStartedBy

id : EString

time : EString

WasEndedBy

id : EString

time : EString

WasDerivedFrom

id : EString

time : EString

HadPrimarySource

WasQuotedFrom

WasRevisionOf

WasAssociatedWith

id : EString

ActedOnBehalfOf

id : EString

WasAttributedTo

id : EString

Bundle

EmptyCollection

Collection

AttributeValue

name : EString

value : EString

[0..*] used

[0..*] entity

[0..*] UsedAttributes

[0..*] EntityAttributes

[0..*] ActivityAttributes

[0..*] wasInformedBy

[0..*] activity

[0..*] WasInformedByAttributes

[0..*] wasGeneratedBy

[0..*] activity

[0..*] WasGeneratedByAttributes

[0..*] wasInvalidatedBy

[0..1] activity

[0..*] WasInvalidatedByAttributes

[0..*] wasStartedBy

[0..*] wasEndedBy

[0..1] entityTrigger

[0..*] WasStartedByAttributes

[0..*] WasEndedByAttributes

[0..*] WasDerivedFromAttributes

[0..1] activity

[0..1] generation

[0..1] usage

[0..*] wasRevisionOf

[0..*] wasQuotedFrom

[0..*] hadPrimarySource

[0..*] actedOnBehalfOf

[0..*] agent

[0..*] wasAttributedTo

[0..*] agent

[0..*] WasAttributedToAttributes

[0..*] ActedOnBehalfOfAttributes

[0..*] wasAssociatedWith

[0..*] agent

[0..*] WasAssociatedWithAttributes

[0..1] activity

[0..1] plan

[0..*] AlternateOf

[0..*] SpecializationOf

[0..*] hadMember

[0..*] activity

[0..*] entity

[0..*] agent

[0..*] wasDerivedFrom

[0..*] entity

[0..*] attributevalue

[0..*] AgentAttributes

Figure 2: Ecore4PROV-DM: Ecore metamodel representing the PROV-DM Data Model.

MDE-Based Graphical Tool for Modeling Data Provenance According to the W3C PROV Standard

145

Figure 3: MDE-based graphical tool for modeling data provenance according to the W3C PROV standard.

nent 4 (Bundles).

In the MQR07 context, evaluators highlighted that

the “Specialization” (specializationOf ) concept in

Component 5, while correctly modeled, is less appar-

ent in the metamodel, with 33.3% ﬁnding it not eas-

ily identiﬁable. Similarly, the “Member” (hadMem-

ber) concept in Component 6 and the “Alternative”

(alternateOf ) concept in Component 5 were identi-

ﬁed as not evident by 26.6% of specialists. However,

the majority of the twenty-two concepts across the six

PROV-DM components were deemed clearly evident

in Ecore4PROV-DM by most evaluators. A compre-

hensive report detailing the outcomes of this evalua-

tion is subject to publication in a forthcoming paper.

4.2 Development of the Graphical Data

Provenance Modeling Tool

The second stage focused on constructing the graph-

ical modeling tool, utilizing the Ecore4PROV-DM

metamodel from the preceding stage. The concrete

syntax of the metamodel was implemented in Eclipse

Sirius, using the Obeo Designer

tool, which was cho-

sen for its seamless integration with EMF, Acceleo,

and Sirius, along with its capacity to switch perspec-

tives during modeling. This choice facilitated the cre-

ation of an intuitive and visually expressive modeling

environment. To ensure adherence to the W3C PROV

standard, style deﬁnitions were carefully integrated,

representing entities as yellow oval nodes, activities

as blue rectangles, and agents as orange “pentagon

houses.” The relationships between these elements

were depicted as oriented edges with labels denoting

the relationship type.

https://www.obeodesigner.com/

4.3 Model-to-Text Transformation

In the third stage, templates were developed within

Eclipse Acceleo for all the elements that make up the

six components of the PROV data model, in order to

allow the provenance model developed in the graph-

ical modelling tool to be converted into PROV nota-

tion (PROV-N) code. The language syntax of each el-

ement followed the PROV-DM documentation guide-

lines (Moreau and Groth, 2013). This stage enabled

the synthesis of a comprehensive provenance model,

leveraging the PROV data model as the underlying

foundation while embracing the expressive power of

the PROV-N notation.

Figure 3 provides an overview of the graphical

provenance modeling tool. The Element Palette (a)

on the right allows the selection and addition of W3C

PROV standard elements (Activity, Agent, and En-

tity) to the graphical model by dragging them to the

Diagram area (b). Clicking on an element in the Di-

agram area or using the Tree Editor (c) enables edit-

ing of its Properties (d) and the addition or modiﬁca-

tion of relationships with other model elements. The

Execute icon (e) triggers the execution of the model-

ing tool, generating corresponding PROV-N code that

represents the graphical model. This code encapsu-

lates the model’s relationships, attributes, and struc-

ture, providing a textual representation of the prove-

nance information encoded in the graphical represen-

tation.

MODELSWARD 2024 - 12th International Conference on Model-Based Software and Systems Engineering

146

5 MODELING DATA

PROVENANCE GRAPHICALLY

In this section, we demonstrate the use of the graph-

ical provenance modeling tool. After modeling each

example, the tool automatically generated the respec-

tive PROV-N code, which was subsequently manually

validated for correctness, ensuring that the tool is gen-

erating valid PROV-N code. For validation, we used

the PROV document validation service provided by

the Open Provenance website

. Due to space limi-

tations, we will refrain from presenting the PROV-N

codes generated by the tool.

5.1 Image Acquisition via Sensors

The ﬁrst example demonstrates the graphical model-

ing of a provenance model that is publicly available

on ProvStore

, a cloud-based provenance repository.

This particular provenance model is associated

with the Urban Observatory project at Newcastle Uni-

versity in the United Kingdom. The Observatory

monitors various urban indicators through streams of

active sensors and CCTV cameras. Due to space

constraints, we have only modeled a portion of this

provenance model within the tool. Speciﬁcally, we

have focused on the bundle that speciﬁes the prove-

nance data collected during the image acquisition pro-

cess from sensors. This model includes two enti-

ties, var:sensor and var:image, along with one activ-

ity, var:acquisition. Each of these model components

has its set of properties and values, detailing the data

collection process. Figure 4 illustrates the graphical

provenance model.

var:sensor

var:acquisition

var:image

uo:value: var:brokerName

uo:value: var:groundHeightAboveSeaLevel

uo:value: var:locationWKT

uo:value: var:sensorCentroidLatitude

uo:value: var:sensorCentroidLongitude

uo:value: var:sensorHeightAboveGround

uo:value: var:sensorName

uo:value: var:thirdParty

uo:value: var:class

uo:value: var:filename

uo:value: var:height

uo:value: var:width

uo:value: var:xmax

uo:value: var:xmin

uo:value: var:ymax

uo:value: var:ymin

uo:value: var:acquisition_date

used

wasGeneratedBy

Figure 4: Graphical provenance model depicting an image

acquisition process using a sensor.

5.2 Graphical Modeling of Provenance

Templates

A provenance template (PROV-Template) (Moreau

et al., 2018) is a PROV document that delineates the

https://openprovenance.org/service/validator.html

https://openprovenance.org/store/documents/497

desired provenance to be generated. Within a prove-

nance template, variables are employed as placehold-

ers for values. Consequently, a provenance tem-

plate serves as a declarative speciﬁcation of the in-

tended provenance to be produced by an applica-

tion. A collection of bindings establishes associ-

ations between variables and corresponding values.

The PROV-Template expansion algorithm, when pro-

vided with a template and a set of bindings, produces

a provenance document in which variables have been

substituted with their corresponding values.

Hence, the ﬁrst step in this methodology involves

designing a “provenance model” that describes the

structure of the provenance that is intended to be gen-

erated or captured. This task is usually done by draft-

ing the provenance model on paper and then translat-

ing the model manually into PROV-N notation. How-

ever, this process is both antiquated and susceptible

to errors. By employing the graphical provenance

modeling tool introduced in this paper, it becomes

possible to interactively design the provenance model

graphically. Once the model is ﬁnalized, generating

its PROV-N code is a straightforward task. This code

is then employed by the expansion algorithm, in con-

junction with the “binding” ﬁle and the designated

database serving as the source of provenance data.

Figure 5 illustrates the provenance model pre-

sented in (Moreau, 2017), describing an arithmetic

operation, for example, the sum of two values re-

sulting in a third value. Thus, the arithmetic oper-

ation type is denoted by the activity var:operation,

which uses two input values, represented by the enti-

ties var:consumed1 and var:consumed2, producing a

result represented by the entity var:produced, which

is triggered by an var:agent.

var:consumed1

var:operation

var:consumed2

var:agent

prov:value: var:consumed_value2

prov:type: var:operation_type

prov:value: var:consumed_value1

var:produced

prov:value: var:produced_type

prov:value: var:produced_value

used used

wasAssociatedWith

wasGeneratedBy

wasDerivedFrom wasDerivedFrom

Figure 5: Graphical provenance model depicting the PROV-

Template for an arithmetic operation.

MDE-Based Graphical Tool for Modeling Data Provenance According to the W3C PROV Standard

147

6 CONCLUDING REMARKS

This paper presented a graphical tool for data prove-

nance modeling based on the W3C PROV standard.

The development involved three key stages: con-

verting the PROV data model (PROV-DM) into an

Ecore metamodel, constructing the graphical tool

with Eclipse Sirius, and utilizing Eclipse Acceleo for

model-to-text transformations to generate PROV-N

code. The Ecore4PROV-DM metamodel underwent

evaluation by ﬁfteen metamodeling experts, resulting

in consensus on its alignment with PROV-DM.

The graphical tool addresses a signiﬁcant gap

by providing a user-friendly interface for creating

expressive data provenance models, enhancing vi-

sualization, and comprehension of provenance rela-

tionships. It allows exporting PROV-N code from

the graphical model, ensuring interoperability and

compatibility with existing provenance representa-

tions. The tool facilitates seamless integration of data

provenance into systems, fostering the adoption of

provenance-aware applications and extending its util-

ity to diverse applications, including provenance in-

formation extraction from databases using the PROV-

Template approach.

While demonstrating the tool’s use in certain sce-

narios, further evaluation is crucial for future re-

search. Usability studies and expert feedback would

enhance functionality, user experience, and effective-

ness. Investigating the tool’s integration with other

provenance management systems and assessing per-

formance in real-world scenarios will contribute to its

practical applicability.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the support pro-

vided by the Instituto Federal Goiano in facilitating

and funding this research.

REFERENCES

Bruel, J.-M., Combemale, B., Guerra, E., J

equel, J.-

M., Kienzle, J., de Lara, J., Mussbacher, G., Syri-

ani, E., and Vangheluwe, H. (2018). Model trans-

formation reuse across metamodels. In Rensink, A.

and S

anchez Cuadrado, J., editors, Theory and Prac-

tice of Model Transformation, pages 92–109, Cham.

Springer International Publishing.

Bucchiarone, A., Cabot, J., Paige, R. F., and Pierantonio, A.

(2020). Grand challenges in model-driven engineer-

ing: an analysis of the state of the research. Software

and Systems Modeling, 19(1):5–13.

Glavic, B. (2021). Data provenance. Foundations and

Trends

 in Databases, 9(3-4):209–441.

Herschel, M., Diestelk

amper, R., and Ben Lahmar, H.

(2017). A survey on provenance: What for? What

form? What from? The VLDB Journal, 26(6):881–

906.

Hu, R., Yan, Z., Ding, W., and Yang, L. T. (2020). A

survey on data provenance in IoT. World Wide Web,

23(2):1441–1463.

Huynh, T. D. (2020). PROV Python - A library for

W3C Provenance Data Model supporting PROV-

JSON, PROV-XML and PROV-O (RDF). Available

online: https://pypi.org/project/prov/.

Huynh, T. D. and Moreau, L. (2015). ProvStore: A Public

Provenance Repository. In Lud

ascher, B. and Plale,

B., editors, Provenance and Annotation of Data and

Processes, pages 275–277, Cham. Springer Interna-

tional Publishing.

Kudo, T. N., Bulc

ao-Neto, R. F., and Vincenzi, A. M. R.

(2020). Metamodel Quality Requirements and Evalu-

ation (MQuaRE). Technical report, Departamento de

Computac¸

ao, UFScar, S

ao Carlos-SP, Brazil. v 2.0.

opez-Fern

andez, J. J., Cuadrado, J. S., Guerra, E., and

de Lara, J. (2015). Example-driven meta-model devel-

opment. Software & Systems Modeling, 14(4):1323–

1347.

Madiot, F. and Paganelli, M. (2015). Eclipse sirius demon-

stration. P&D@ MoDELS, 1554:9–11.

Moreau, L. (2016). ProvToolbox - Java library to cre-

ate and convert W3C PROV data model representa-

tions. Available online: https://lucmoreau.github.io/

ProvToolbox/.

Moreau, L. (2017). PROV-Template: A Quick Start. Avail-

able online: https://lucmoreau.wordpress.com/2017/

03/30/prov-template-a-quick-start.

Moreau, L., Batlajery, B. V., Huynh, T. D., Michaelides,

D., and Packer, H. (2018). A templating system to

generate provenance. IEEE Transactions on Software

Engineering, 44(2):103–121.

Moreau, L. and Groth, P. (2013). Provenance: An Introduc-

tion to PROV. Springer International Publishing.

Moreau, L., Missier, P., Belhajjame, K., B’Far, R., Cheney,

J., Coppens, S., Cresswell, S., Gil, Y., Groth, P., Lebo,

G. K. T., McCusker, J., Miles, S., Myers, J., and Sa-

hoo, S. (2013a). PROV-DM: The PROV Data Model.

Available online: https://www.w3.org/TR/prov-dm/.

Moreau, L., Missier, P., Cheney, J., and Soiland-Reyes, S.

(2013b). PROV-N: The Provenance Notation. Avail-

able online: https://www.w3.org/TR/prov-n/.

erez, B., Rubio, J., and S

aenz-Ad

an, C. (2018). A system-

atic review of provenance systems. Knowledge and

Information Systems, 57(3):495–543.

Rodrigues da Silva, A. (2015). Model-driven engineering:

A survey supported by the uniﬁed conceptual model.

Computer Languages, Systems & Structures, 43:139–

155.

Steinberg, D., Budinsky, F., Merks, E., and Paternostro, M.

(2008). EMF: Eclipse Modeling Framework. Pearson

Education.

MODELSWARD 2024 - 12th International Conference on Model-Based Software and Systems Engineering

148