Using Context Elements and Data Provenance to Support Reuse in

Scientiﬁc Software Ecosystem Platform

Lenita M. Ambr

osio, Jos

e Maria N. David, Regina Braga,

Fernanda Campos, Victor Str

oele and Marco Ant

onio Ara

ujo

Department of Computer Science, Federal University of Juiz de Fora (UFJF), Juiz de Fora, MG, Brazil

{marco.araujo, victor.stroele}@ice.ufjf.br

Keywords:

Contextual Elements, Provenance, Scientiﬁc Experiments, E-Science.

Abstract:

[Background] Managing contextual elements and provenance information plays a key role in the context of

scientiﬁc experiments. Currently the scientiﬁc experimentation process requires support for collaborative and

distributed activities. Detailed logging of the steps to produce results, as well as the environment context in-

formation could allow scientists to reuse these results in future experiments and reuse the experiment or parts

of it in another context. [Objectives] The goal of this paper is to present a provenance and context metadata

management approach that support researchers to reuse experiments in a collaborative and distributed plat-

form. [Method] First, the context and provenance management life cycle phases were analyzed, considering

existing models. Then it was proposed a conceptual framework to support the analysis of contextual elements

and provenance data of scientiﬁc experiments. An ontology capable of extracting implicit knowledge in this

domain was speciﬁed. This approach was implemented in a scientiﬁc ecosystem platform. [Results] An initial

evaluation shown evidences that this architecture is able to help researchers during the reuse and reproduction

of scientiﬁc experiments. [Conclusions] Context elements and data provenance, associated with inference

mechanisms, can be used to support the reuse in scientiﬁc experimentation process.

1 INTRODUCTION

Over the last decades, science has explored new pos-

sibilities for scientiﬁc experimentation. Complex

phenomena have been simulated by supercomputers

using computational tools. Nowadays, a new para-

digm arises to support scientiﬁc experiments, which

are focused on the open and data-intensive science.

In this paradigm, the information is the main pro-

duct. It is also critical that scientists can share in-

formation with other members of the community, as

well as reuse data from their colleagues. However,

the reuse of knowledge gained in experiments, produ-

ced by third parties, is still a challenge. Federer et al.

(2015) list several reasons why researchers still do not

share data. In addition, even if the data is shared, it

needs to be interpreted appropriately by different re-

search groups. The proper interpretation of the shared

data set requires that this set be complemented by des-

criptive metadata (Missier et al., 2010).

This collaborative and distributed scientiﬁc expe-

rimentation scenario also requires that social and or-

ganizational aspects be considered, since the know-

ledge about the way experiments are performed can

be tacit and often remains with the researcher. Thus,

storing and retrieving contextual information during

the experimentation process may be critical if its acti-

vities are performed in order to be reproducible and

reusable (Mayer et al., 2014).

Considering these challenges, information about

the context and provenance of scientiﬁc experiments

plays a key role. Provenance information describes

the origin, derivation, ownership, and history of the

data (Lim et al., 2010). Context is a complex descrip-

tion of shared knowledge about physical, social, his-

torical or other circumstances within which an action

or an event occurs (Rittenbruch, 2002). In scientiﬁc

experimentation domain, we consider provenance in-

formation as a kind of contextual element that descri-

bes information in the past. Thus, this information is

fundamental so that researchers can understand, re-

produce, examine, and audit the results previously

obtained by experiments, as well as reuse an expe-

riment or parts of it.

Provenance management has been widely discus-

sed in the scientiﬁc community (Simmhan et al.,

Ambrósio, L., David, J., Braga, R., Campos, F., Ströele, V. and Araújo, M.

Using Context Elements and Data Provenance to Support Reuse in Scientiﬁc Software Ecosystem Platform.

DOI: 10.5220/0006676302550262

In Proceedings of the 20th International Conference on Enterprise Information Systems (ICEIS 2018), pages 255-262

ISBN: 978-989-758-298-1

255

2005; Lim et al., 2010). In some Scientiﬁc Work-

ﬂow Management Systems (SWfMS), such as Kep-

ler

, Taverna

and VisTrails

, provenance information

is automatically captured. However, in general, their

proprietary models make it difﬁcult to share infor-

mation. Other provenance approaches (Costa et al.,

2014; Cuevas-Vicentt

ın et al., 2014) cover only the

capture of information from scientiﬁc workﬂows in

isolation to the experiment. In each of these appro-

aches, information may be available at a speciﬁc ab-

straction level, which may or may not be appropri-

ate for the type of analysis required in the experiment

context (Missier, 2016).

On the other hand, the use of contextual informa-

tion in scientiﬁc experimentation is an incipient topic.

ezillon (2011), for example, presents a contextual

approach to support researchers ﬁnding correct scien-

tiﬁc workﬂows in the repository. Mayer et al. (2014)

suggest an ontology-based model to describe scien-

tiﬁc experiments facilitating their reuse and reprodu-

cibility. These researchers describe speciﬁc approa-

ches, but do not provide guidelines that can support

context management in a collaborative and distribu-

ted platform.

In the software development scenario, one of the

approaches used to deal with the need of collabora-

tion and distribution in a heterogeneous environment

is Software Ecosystems (Manikas, 2016). In the con-

text of this work, a software ecosystem consists of

relationships among suppliers of scientiﬁc software,

research institutes, development agencies, funding in-

stitutions and stakeholders to provide and reuse rese-

arch result, supported by a technological infrastruc-

ture (Freitas et al., 2015; Manikas, 2016).

In this vein, Freitas et al. (2015) created the E-

SECO (E-Science Software ECOsystem) platform.

This platform can manage and support all stages of

scientiﬁc experiment life cycle. However, it does not

support provenance and context management to help

in the reuse of scientiﬁc experiments.

Therefore, considering the E-SECO platform this

work proposes an architecture for the management of

provenance and context information that helps resear-

chers to understand scientiﬁc experiments and reuse

them. In order to reach this objective, this work is ba-

sed on provenance and context models proposed by

Missier (2016) and Br

ezillon et al. (2004) respecti-

vely. The main contribution of this work is the spe-

ciﬁcation of a metadata management architecture, na-

med ContextProv, which aims to manage provenance

and context information of scientiﬁc experiments in a

https://kepler-project.org

http://www.taverna.org.uk

https://www.vistrails.org

software ecosystem platform.

This article is organized in ﬁve sections besi-

des Introduction. Section 2 presents Related Works.

Section 3describes the proposed architecture. Section

4 presents an initial evaluation of the solution con-

sidering the ContextProv architecture use in the E-

SECO Platform. Section 5 presents the Final Con-

siderations and Future Works.

2 RELATED WORKS

The management of provenance data as well as con-

text information is not recent in the literature. Howe-

ver, the existing works do not associate context and

provenance concepts to experiments reuse in scienti-

ﬁc software ecosystems platforms.

ProvSearch (Costa et al., 2014) proposes a pro-

venance management architecture for experiments in

distributed environments. It combines distributed

workﬂow management techniques with provenance

data management. It also allows provenance data to

be captured, stored and queried at run-time. In this

architecture, data is fragmented into multiple reposi-

tories of provenance in the cloud which can be acces-

sed by different SWfMSs. PBase (Cuevas-Vicentt

ın

et al., 2014) is a scientiﬁc workﬂow provenance re-

pository that uses the ProvONE ontology (Cuevas-

Vicentt

ın et al., 2014), allowing the storage, analysis

and replication of scientiﬁc experiments.

ezillon (2011) presents an approach that uses

contextual graphs to support researchers when reu-

sing scientiﬁc workﬂows. This approach helps re-

searchers to ﬁnd a workﬂow through a long process

of contextualization (identifying the published work-

ﬂow that has a context close to the desired one). In

addition, it supports the decontextualization, which

extracts parts of the workﬂow that can be reused

in a relatively generic way, and the recontextualiza-

tion, which develops workﬂow instances adapted to

new contexts. TIMBUS context model (Mayer et al.,

2014) is a model for the description of scientiﬁc ex-

periments focused speciﬁcally on the technical infra-

structure used as the basis for the experiment. It aims

to preserve the processes, the architectural principles,

and the core ontologies to extend the experiment, al-

lowing its reuse and reproducibility.

There are other approaches that deal with pro-

venance or context management in scientiﬁc experi-

ments. However, these approaches deal with contex-

tual or provenance information in isolation. They do

not consider both concepts in a distributed and colla-

borative context, as we do in our work. In addition,

they treat speciﬁc problems, and consider provenance

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

256

and context concepts focusing only on the workﬂow

and its results. They do not address the whole process

of planning, design, and execution of an experiment

and its related workﬂows. As a result, these approa-

ches are not able to support activities throughout the

scientiﬁc experimentation life cycle, using a scientiﬁc

software ecosystem platform.

3 PROVENANCE AND CONTEXT

IN SCIENTIFIC EXPERIMENTS

In this section, E-SECO platform and the proposed

approach for provenance and context management in

scientiﬁc experiments are presented.

3.1 E-SECO Ecosystem Platform

E-SECO is a software ecosystem platform developed

to support activities carried out during the life cycle of

scientiﬁc experiments. The key modules of this plat-

form have already been developed and evaluated in

e-Science domain, and are illustrated in Figure 1. E-

SECO Development Environment is a web component

where E-SECO code is available, as open source

As a result, the developer community can contri-

bute through software maintenance and evolution. E-

SECO relies on a Peer-to-Peer network where diffe-

rent E-SECO nodes can communicate. The ecosy-

stem is made up of artifacts provided by different no-

des situated in different institutions, APIs that help the

scientiﬁc workﬂow development in its different steps

and the open source development environment.

Figure 1: E-SECO Platform Architecture Overview.

The visualization module of E-SECO platform,

named Multi-Layer Visualization, supports the ex-

traction and analysis of the relationships that are es-

tablished in scientiﬁc social networks. E-SECO plat-

form is integrated with External Applications through

an Integration Layer. Due to space restrictions, E-

SECO platform is not discussed in depth. A detailed

http://pgcc.github.io/plscience/

presentation of this platform was done by Freitas et al.

(2015), Sirqueira et al. (2016).

In order to support reuse during the experimen-

tation process, the ContextProv architecture extends

provenance management and adds the management of

contextual elements.

3.2 ContextProv Architecture

The veriﬁcation, reproduction, and reuse of scientiﬁc

experiments are key activities to support researchers

conducting their experiments in a shorten time, and

with higher quality. However, these activities are not

trivial. Experiments change and evolve over time ac-

cording to their contexts. As new results emerge, re-

search may follow different approaches and new tasks

may arise. As a consequence, planning, modiﬁcation,

or adaptation of the execution process, or even new

third-parties’ resources are required. Moreover, in a

software ecosystem platform, a set of new require-

ments emerge that need to be fulﬁlled. As example

we can mention, service composition, interoperabi-

lity and extensibility support. E-SECO platform also

aims to fulﬁll these requirements.

In order to reuse input data produced by a resear-

cher in other context, it is not enough that these data

are published on a shared platform. The adequate in-

terpretation requires that these data are complemen-

ted by descriptive metadata (Missier et al., 2010).

ContextProv architecture is intended to perform me-

tadata management of provenance and context over

the experiment life cycle, provinding relevant infor-

mation to support researchers during the experimen-

tation process.

In previous works, the experiment life cycle has

already been extended (Freitas et al., 2015; Sirqueira

et al., 2016), but they do not consider contextual and

provenance elements. Figure 2 illustrates the experi-

ment life cycle, and highlights some of the contex-

tual elements and provenance information captured

throughout this cycle. During the Investigation of the

Problem and Prototyping phases, the deﬁnition of the

scope of the research takes place. In addition, servi-

ces and workﬂows to support the experiment are de-

veloped and/or reused. In these phases, the elements

of the development context and the prospective pro-

venance information are captured. This information

represents an abstract speciﬁcation of the experiment

as a guideline for the derivation of future data.

During the Execution of the Experiment and Pu-

blication of Results phases, the experiment is carried

out in a controlled manner and the results as well as

data related to the experimentation process are sto-

red and published. In these phases, context elements

Using Context Elements and Data Provenance to Support Reuse in Scientiﬁc Software Ecosystem Platform

257

Figure 2: ContextProv Overview.

of the execution phase and the retrospective prove-

nance are captured. They represent, for example,

which tasks were performed and how data artifacts

were derived. This information helps to verify the re-

sults obtained by the experiment.

Throughout the experiment life cycle, contextual

and provenance information is captured and stored in

a repository. Subsequently, data in this repository are

processed by inference machines aimed to extract im-

plicit knowledge about the experiment. Through a vi-

sualization mechanism, this information also genera-

tes knowledge to support the researchers in future ex-

periment processes. This process of capture, storage,

inference and visualization of provenance and context

data is done by Prov-SE and Context-SE modules.

3.2.1 Prov-SE Module

It has as its main goal the management of provenance

metadata over the experiment life cycle. It also al-

lows the capture and query of provenance data to sup-

port the researcher during the experimentation pro-

cess. In order to describe the steps to manage prove-

nance, Missier (2016) proposed a framework for the

provenance life cycle. This framework illustrates the

main phases of data provenance until it can be viewed

or analyzed. Prov-SE was built based on Missier’s

model, and includes the capture, storage, inference,

query, share and visualization phases.

Initially, provenance information is generated

from the experiment speciﬁcation, during the plan-

ning and modeling of the associated workﬂows. In the

execution phase of workﬂows, data is collected by a

Web Service, which is part of the E-SECO ProVersion

architecture (Freitas et al., 2015). This service captu-

res information about the experiments execution, such

as: start and end time of the execution, input and out-

put data of each task performed by the workﬂows, and

ﬁnal results.

Some SWfMS, such as Kepler, exports prove-

nance information. Then this data can be imported

into E-SECO platform, enriching the information al-

ready captured. The data captured are stored in distri-

buted repositories, according to the remote execution

of the workﬂows. They are modeled according to the

ProvONE conceptual data model (Cuevas-Vicentt

ın

et al., 2014). Thus, data are stored according to a

standard format, aiming to facilitate their interpreta-

tion and interoperability across other systems. To deal

with a large volume of data, E-SECO platform imple-

ments a peer-to-peer (P2P) network. Each node of

the network has an E-SECO data repository, which

allows the storage of these data in a decentralized but

uniform way.

In order to extract implicit knowledge from the

provenance data, an ontology was developed, named

Prov-SE-O, which is based on the ProvONE model.

The main purpose of the original ProvONE ontology

is to capture data from scientiﬁc experiments rela-

ted to a speciﬁc workﬂow, its derivations and sub-

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

258

workﬂows. An explicit concern with the distributed

nature of the scientiﬁc experimentation process is not

part of the original ProvONE ontology. So, the dis-

tributed nature of scientiﬁc experimentation process

is considered by Prov-SE-O ontology. Prov-SE-O al-

lows the modeling of scientiﬁc experiment as a whole,

including the related distributed workﬂows, captu-

ring important information related to its distributed

and collaborative nature. This ontology is detailed in

(Ambr

osio et al., 2017).

The data captured by the Prov-SE module is lo-

aded into the ontology, and through inference algo-

rithms, implicit knowledge can be derived. For the de-

rivation of this knowledge, Prov-SE-O ontology uses

Property Chains

and SWRL (Semantic Web Rule

Language) rules (Horrocks et al., 2004).

In addition to the P2P network, this module pro-

vides a semantically annotated RESTful web service

to facilitate data sharing. The P2P network allows the

data sharing between different instances of E-SECO

platform. In the other hand, the Web Service allows

that this data can also be shared with other platforms

or external services.

This module also uses a visualization tool based

on the PROV model. In this step, the stored data are

loaded into the ontology and the visualization is gene-

rated from the developed tool (Oliveira et al., 2017).

Information about the provenance of scientiﬁc ex-

periments can be considered as a kind of contex-

tual information. However, they are not sufﬁcient to

support researchers during the experiment execution.

Dealing with a collaborative and distributed activity,

researchers need to know the results of the indivi-

dual work of the group participants. Otherwise, there

will be no collaboration, but set of isolated activities

(Br

ezillon et al., 2004). Group work needs explicit

context management. Hence, the ContextProv archi-

tecture encompasses Context-SE module.

3.2.2 Context-SE Module

This module is based on the framework proposed by

ezillon et al. (2004) which presents groupware me-

chanisms associated with explicit context represen-

tation. Similar to the model used to support pro-

venance, this module supports generation, capture,

storage, awareness, interpretation and visualization

stages.

Context model was developed in order to deter-

mine what part of the contextual information is rele-

vant to scientiﬁc experimentation. This model con-

Property Chains appeared in OWL 2 and works by sor-

ting objects, where it allows transitivity between multiple

properties.

siders the absolute / relative space and time dimen-

sions, context history, subject, and user proﬁle (Bol-

chini et al., 2007). As a result, it aims to make con-

text management more efﬁcient by storing only rele-

vant information. Annotating all the critical details

of an experiment in a laboratory notebook is a stan-

dard scientiﬁc procedure, especially in the experimen-

tal sciences. The key issue in e-Science is that the

number and granularity of critical details are high, and

identifying them thoroughly is a challenge, and wri-

ting all of them is time-consuming. In this way, the

Context-SE module has a conceptual framework to

identify and classify common contextual elements in

a collaborative environment of scientiﬁc experimen-

tation.

This framework is an extension of the conceptual

framework proposed by Rosa et al. (2003) which con-

siders the relevant elements for context analysis in

groupware applications. Context-SE framework mo-

deled these elements in ﬁve context categories.

The ﬁrst category refers to information about

group members. This is information about resear-

chers and research groups to which they belong. The

second category concerns information about schedu-

led tasks. In scientiﬁc experimentation domain it is

related to the planning of the experiment and is cha-

racterized by the tasks to be performed by the group

until the conclusion of the experiment.

The third category concerns the relationship bet-

ween group members and scheduled tasks. It relates

each researcher or research group to the interactions

in which they are involved. This category is divided

into two types of contexts: interaction context (infor-

mation representing the actions that occurred during

the experiment execution) and the planning context

(information about the project execution plan).

The fourth category brings together information

about the environment. It covers both organizational

issues and the technological environment, that is, all

information outside the experiment, but within the or-

ganization that can affect the way the tasks are per-

formed. Finally, the ﬁfth category gathers all the in-

formation about the completed tasks. Its purpose is

to provide basic information about the lessons lear-

ned, whether from the same group or from similar

tasks carried out by other groups. It should there-

fore include all contextual and provenance informa-

tion about previous experiments.

Ambr

osio et al. (2017) present a detailed descrip-

tion of this framework. Following this guidelines, E-

SECO platform captures information, such as:

1. Researchers and Research Groups. This infor-

mation helps other researchers identifying who

are involved in the experiment. So, it is possi-

Using Context Elements and Data Provenance to Support Reuse in Scientiﬁc Software Ecosystem Platform

259

ble to contact these researchers to collaborate on

a particular experiment.

2. Planning the Experiment. This information re-

fers to the experiment, the workﬂows and the tasks

to be performed, which contribute for the experi-

ment to be reused by other researchers, in a new

context.

3. Relationship between Tasks and Researchers.

This information allows credits to be given to the

authors, and that they are responsible or questio-

ned for any errors that occurred during its execu-

tion.

4. Environment. Information about used technolo-

gies, SWfMS, and external services are also es-

sential for the reproducibility of the experiment.

5. Tasks Completed and Provenance. Information

about the experiments, workﬂows, and tasks exe-

cuted are stored, as well as the provenance of

them. Thus, it is possible to identify all the pro-

cesses undergone by an artefact, until the end of

the experiment, as well as to identify the work-

ﬂow and the experiment that gave origin to this

artefact, and the involved researchers.

This contextual information can be informed by

the researcher, but must be primarily obtained through

integration with other platforms, such as Mendeley,

which allow the extraction of information about rese-

archers, research groups and institutions.

All contextual information captured by the plat-

form are stored in a repository. In addition, this repo-

sitory data is connected with related provenance in-

formation, which describes the completed tasks, and

their associated contexts elements. This information

can also be processed by the ontology, allowing the

extraction of implicit knowledge.

4 ContextProv IN ACTION

This section presents a feasibility study, with the aim

of detailing the reproducibility support, from the use

of the ContextProv architecture, in the E-SECO plat-

form. This study is composed of three phases, which

detail the activities of experimentation, reuse and re-

producibility. During the analysis of this scenario, we

tried to answer questions that arise when reusing or

reproducing an experiment. These questions are:

Question 1. What is the process responsible for

constructing a given result?

Question 2. Is there another similar experiment?

Question 3. Why two similar experiments yield

different results?

4.1 Experimentation

An oncologist researching the mutations of the

ATRX

protein in human tumors wishes to keep up

to date with published research related to this protein.

His research group uses the E-SECO platform to fa-

cilitate collaboration between researchers. Thus, this

scientist is part of an E-SECO research community

that explores the use of this protein and its effects.

This scientist receives information that the Nu-

clear Protein Database

(NPD), provides a Web Ser-

vice that searches for information about a given pro-

tein, including links to papers published in PubMed

Entrez Protein

bases, among others. In addition, the

E-SECO platform recommends the use of a work-

ﬂow, available at Kepler website and linked to E-

SECO repository, which searches the NPD database

and counts the number of published articles on this

protein. This workﬂow, named PapersCount, can be

executed occasionally to update the information. The

workﬂow version has speciﬁc services that let the use

of the ContextProv architecture to allow the capture

of contextual information during its execution.

Thus, from the execution of the PapersCount

workﬂow, in the context of the ”ATRX protein” ex-

periment, provenance and context information related

to each of the workﬂow executions is generated. This

data is stored in one of the E-SECO repositories.

4.2 Reuse

Another scientist, biochemist specialized in studying

proteins present in the human body, is part of another

research group, related to proteins. This scientist also

studies the ATRX protein, and needs to ﬁnd scientiﬁc

articles related to this protein. From an initial search

conducted by this scientist, the E-SECO platform re-

turns information about the ATRX protein. The bio-

chemist found that there are currently 52 articles on

the ATRX protein. Thus, this scientist questioned:

(Q1) What process is responsible for this result?

Analyzing the information about result’s prove-

nance, available at E-SECO platform he can see that

it was obtained through the execution of the Pa-

persCount workﬂow. Based on this information, the

scientist would like to reuse the PapersCount work-

ﬂow by restructuring only the last task. Instead of

returning the number of articles found, would return

the links to these articles.

Transcriptional regulator ATRX is a protein that in hu-

mans is encoded by the ATRX gene.

http://npd.hgu.mrc.ac.uk

https://www.ncbi.nlm.nih.gov/pubmed

http://www.ncbi.nlm.nih.gov/entrez

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

260

However, before reusing this workﬂow, the bio-

chemist wants to be sure of the reliability of the re-

sults. The scientist accessed the provenance informa-

tion of the PapersCount workﬂow (Figure 3). Based

on this information, he took notice about the origin

of the workﬂow, the qualiﬁcations of the researcher

who used it and his work institution. Knowing that

the authorship of the workﬂow (Kepler development

group) and the scientist who used it as well as his in-

stitution, are recognized by the scientiﬁc community,

the biochemist considered the PapersCount workﬂow

as reliable. So, he decided to reuse it in his research.

In this way, a new version of the workﬂow, called Pa-

persLinks, has been created. In this version, the last

task has been modiﬁed so that its output is a list of

links to the articles found.

Figure 3: PapersCount Provenance – presented in Prot

4.3 Reproducibility

Later, a biologist, a protein researcher connected to

the E-SECO platform, needs to make a comparison

between the ATRX and ATR proteins. As a ﬁrst step,

he wants to check each one has a greater number of

published articles. During the prototyping of his ex-

periment on the E-SECO platform, this scientist con-

ducted a workﬂow search, where he found the Pa-

persLinks workﬂow. Analyzing this workﬂow, he rea-

lized that its output was not exactly what he required,

as he needed to count the number of articles. Then he

wondered: (Q2) Are these experiments similar?

To answer this question, the scientist searched

for similar experiments on the E-SECO platform.

Through the execution of inferences of the Prov-SE-

O ontology, the E-SECO platform shown that the

PapersCount experiment seems to be similar to Pa-

persLinks because they have common tasks, and use

the same input information and the same Web Ser-

vice. Through the use of the PapersCount workﬂow,

this scientist realized that it returns exactly what he

needs. As a result, he decided to replicate the expe-

riment that used PapersLinks, for the two proteins he

wants to compare, ATRX and ATR.

When reproducing the experiment for the protein

ATRX, the scientist obtained 63 articles. Comparing

this result with that obtained in the previous experi-

ment (52 articles), he realized that the results were

different. This divergence brings doubt on the relia-

bility of this experiment. So, he need to know: (Q3)

Why did the experiments yield different results?

By analyzing the contextual information captu-

red by the ContextProv architecture, this scientist

can compare workﬂow versions, tasks performed, the

Web Service used and its version, the SWfMS used,

the input and output values of the workﬂow, the scien-

tist who is responsible for the experiment, and the

date of execution. Observing this information, he re-

alized that there was no signiﬁcant change in the con-

text of the two experiments, except that the ﬁrst ex-

periment had been executed more than a year. In this

way, the scientist realized that there are evidences that

the difference in results is due to the new articles pu-

blished during this period.

So, this scenario shows that the ContextProv ar-

chitecture can possibly help researchers to answer

previously raised questions and others, which usually

arise during the reuse and reproduction of scientiﬁc

experiments. In addition, it is also worth noting that

information about provenance and context is essential

in this process, and that the use of inference is capa-

ble of deriving implicit information that would not be

used without the support of ContextProv.

5 CONCLUSIONS

This work presented an architecture to support the

management of provenance and context metadata in

scientiﬁc experiments on E-SECO ecosystem plat-

form. This architecture aims to support researchers

in the understanding and reuse of scientiﬁc experi-

ments in collaborative and distributed environments.

For this purpose, it captures relevant context informa-

tion and data provenance of the experiments, proces-

ses this data in an ontology based on ProvONE model,

and thus succeeds in extracting implicit knowledge.

This information can be queried and visualized in a

standardized way through the ecosystem platform. As

a result, it can facilitate its interpretation and provides

knowledge to the researchers.

Regarding the limitations of this research, we can

point out that the interface of the platform needs to be

Using Context Elements and Data Provenance to Support Reuse in Scientiﬁc Software Ecosystem Platform

261

evaluated in order to implement improvements for the

scientist. For this purpose, we intend to advance expe-

rimental studies, evaluating the operation of this plat-

form integrated to third-parties’ scientiﬁc software

ecosystem platform. Furthermore, extensions to other

SWfMS databases that have proprietary provenance

models also need to be addressed. Results show that

the context elements and provenance data could be

used to support the reuse of scientiﬁc experiments in

a collaborative and distributed environment, but they

cannot be generalized. Experiments need to be car-

ried out considering the real-world contexts the de-

sign decisions of scientists and developers. In the fu-

ture works, we intend to carry out a formal evaluation

of ContextProv architecture through a Case Study at

an Agricultural Research Corporation that conducts

experiments related to feed efﬁciency in dairy cattle.

REFERENCES

Ambr

osio, L. M., David, J. M. N., Braga, R., Str

oele, V.,

Campos, F., and Ara

ujo, M. A. (2017). Prov-se-o: a

provenance ontology to support scientists in scientiﬁc

experimentation process: Wip. In Proceedings of the

12th International Workshop on Software Engineering

for Science, pages 15–21. IEEE Press.

Ambr

osio, L. M., David, J. M. N., Braga, R., Str

oele, V.,

Campos, F., and Ara

ujo, M. A. (2017). Context-SE:

Conceptual framework to analyse context and prove-

nance in scientiﬁc experiments. In 14th Brazilian

Symposium in Collaborative Systems, pages 1372–

1386. CSBC.

Bolchini, C., Curino, C. A., Quintarelli, E., Schreiber, F. A.,

and Tanca, L. (2007). A data-oriented survey of con-

text models. ACM Sigmod Record, pages 19–26.

ezillon, P. (2011). Contextualization of scientiﬁc work-

ﬂows. In International and Interdisciplinary Confe-

rence on Modeling and Using Context, pages 40–53.

Springer.

ezillon, P., Borges, M. R., Pino, J. A., and Pomerol, J.-C.

(2004). Context-based awareness in group work. In

FLAIRS Conference, pages 575–580.

Costa, F., de Oliveira, D., and Mattoso, M. (2014). Towards

an adaptive and distributed architecture for managing

workﬂow provenance data. In Proceedings of the 2014

IEEE 10th International Conference on e-Science, pa-

ges 79–82.

Cuevas-Vicentt

ın, V., Kianmajd, P., Lud

ascher, B., Missier,

P., Chirigati, F., Wei, Y., Koop, D., and Dey, S. (2014).

The PBase scientiﬁc workﬂow provenance repository.

International Journal of Digital Curation, pages 28–

38.

Federer, L. M., Lu, Y.-L., Joubert, D. J., Welsh, J., and

Brandys, B. (2015). Biomedical data sharing and

reuse: Attitudes and practices of clinical and scientiﬁc

research staff. PLOS ONE, 10(6):1–17.

Freitas, V., David, J. M., Braga, R., and Campos, F. (2015).

An architecture for scientiﬁc software ecosystem. In

9th Workshop on Distributed Software Development,

Software Ecosystems and Systems-of-Systems (WDES

2015), pages 41–48. (in portuguese).

Horrocks, I., Patel-Schneider, P. F., Boley, H., Tabet, S.,

Grosof, B., Dean, M., et al. (2004). Swrl: A semantic

web rule language combining owl and ruleml. W3C

Member submission, page 79.

Lim, C., Lu, S., Chebotko, A., and Fotouhi, F. (2010).

Prospective and retrospective provenance collection in

scientiﬁc workﬂow environments. In Services Compu-

ting (SCC), 2010 IEEE International Conference on,

pages 449–456.

Manikas, K. (2016). Revisiting software ecosystems rese-

arch: A longitudinal literature study. Journal of Sys-

tems and Software, 117:84–103.

Mayer, R., Miksa, T., and Rauber, A. (2014). Ontologies for

describing the context of scientiﬁc experiment proces-

ses. In e-Science, 2014 IEEE 10th International Con-

ference on, pages 153–160. IEEE.

Missier, P. (2016). The Lifecycle of Provenance Metadata

and Its Associated Challenges and Opportunities, pa-

ges 127–137. Springer International Publishing.

Missier, P., Lud

ascher, B., Bowers, S., Dey, S., Sarkar, A.,

Shrestha, B., Altintas, I., Anand, M. K., and Goble, C.

(2010). Linking multiple workﬂow provenance traces

for interoperable collaborative science. In Workﬂows

in Support of Large-Scale Science (WORKS), 2010 5th

Workshop on, pages 1–8.

Oliveira, W., Ambr

osio, L. M., Braga, R., Str

oele, V., Da-

vid, J. M., and Campos, F. (2017). A framework for

provenance analysis and visualization. Procedia Com-

puter Science, 108(Supplement C):1592 – 1601. Inter-

national Conference on Computational Science, ICCS

2017, 12-14 June 2017, Zurich, Switzerland.

Rittenbruch, M. (2002). Atmosphere: a framework for con-

textual awareness. International Journal of Human-

Computer Interaction, pages 159–180.

Rosa, M. G., Borges, M. R., and Santoro, F. M. (2003). A

conceptual framework for analyzing the use of context

in groupware. In International Conference on Colla-

boration and Technology, pages 300–313. Springer.

Simmhan, Y. L., Plale, B., and Gannon, D. (2005). A survey

of data provenance in e-science. ACM Sigmod Record,

pages 31–36.

Sirqueira, T. F., Dalpra, H. L., Braga, R., Ara

ujo, M. A.,

David, J. M. N., and Campos, F. (2016). E-SECO

proversion: Manutenc¸

ao e evoluc¸

ao de experimentos

cient

ıﬁcos. In BreSci - 10

Brazilian e-Science Works-

hop, pages 253–260. CSBC.

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

262