There has been work to provide preservation mechanisms that aim to capture the results of computational experiments. However, these preservation environments concern themselves with preserving the computational replication of numeric results rather than context of the calculation that is demanded for scienti cally reproducible and extensible results. This is insu cient to help researches to comprehend preserved experiments and computational artifacts since su cient documentation is lacking so understanding requires face to face meetings with their group members and external collaborators and may likely be lost. Computational experiments need to be described in both human and computer readable fashion. This paper aims at providing an ontology design pattern that conceptualizes any computational experiment artifacts and is aligned with other existing vocabularies such as the well-known W3C PROV vocabulary.
As computational work becomes more and more integral to scienti c research, an e cient approach to preserve and share experiments has become a concern required by the reproducibility of the scienti c process. Although a variety of technologies have been created to capture computational experiments, it is still a struggle for scientists to understand the preserved experiment without a detailed documentation.
In this paper, we present an early work towards building an ontology called
Smart Container (SC) that begins with conceptualizing computational
experiments from the perspective of computational environments and activities within
those environments as provided by the Docker Linux container framework1 as a
preservation tool. Docker is a lightweight virtualization platform that has several
properties such as providing versioned le system, a modular design for
distribution of software components as well as a sustainable community that make
it attractive as a preservation tool. One community of collaborators at CERN2
is already exploring Docker to preserve high energy physics experiments. Our
approach is to eventually provide a mechanism that captures the additional
provenance of computational experiments in a machine readable approach using
the W3C standard RDF data model3 that has been shown to aid in
contextualization of scienti c experiment.[
We propose the construction of a \Smart Container Ontology" using a modular
approach by systematic alignment of concepts present in Docker as a
computational environment where computational activities occur by a modular approach
reusing vocabulary terms where possible to contextualize those concepts and
creation. This pattern based approach is believed to facilitate some control over
unintended consequences of large ontologies and entailment inconsistencies
introduced by their logic descriptions.[
PROV-O is a W3C recommendation that describes \information about
entities, activities, and people involved in producing a piece of data or thing, which
can be used to form assessments about its quality, reliability or trustworthiness"7
and facilitates the exchange of this information between di erent systems and
under di erent contexts. PROV-O has been demonstrated to have reasonable
exibility and has been shown to have promise for enhancing alignment between
di erent ontologies.[
5 http://cos.ontoware.org/cso# 6 http://descartes-core.org/ontologies/activity/1.0/ActivityPattern# 7 http://www.w3.org/TR/2013/NOTE-prov-overview-20130430/ 8 http://ontologydesignpatterns.org/wiki/Submissions:An_Ontology_Design_
Pattern_for_Activity_Reasoning
process for ontology engineering9, describes the common core concept for
reasoning about activities.[
A Proposed Pattern
A Docker based execution engine can be divided into at least two types of
activities: provisioning activities that create an appropriate environment for
a computational activity to occur that directly produce scienti c
observations. Provisioning activities include necessary procedures, such as prepare OS,
install software, install software libraries, and download data, contributing to
the \computational environment" for a scienti c computation. Computational
activities are actual experiment steps and work ows for a scienti c investigation
which involving generating and analyzing research data. We propose aligning
these activity concepts with prov:Activity which represents "something that
occurs over a period of time and acts upon or with entities.[
In Fig.1. we present a concept map of the application of proposed pattern with respect to Docker. The class sc:runningContainer, representing a Docker container, is a specialization of prov:Activity and act:Activity. It has a requirement from ACT that is a sc:RequiredImage, a subclass of prov:Entity and act:Requirement. The outcome class from ACT is represented by sc:OutcomeImage, a subclass of prov:Entity and act:Outcome. A sc:SoftwareAgent is a direct subclass of prov:SoftwareAgent standing for the software executes activities on behalf of the human user. In Fig.2. we present a concept map of the specialization of the activity pattern. The classes sc:provisonActivity and sc:computationalActivity are a subclass of sc:runningContainer. Provisioning activities are in uenced by sc:provisionPlan which subclass prov:Plan. A sc:computationalActivity is also in uenced by a sc:work owPlan10. Provisioning activities produce a computational environment that is a requirement for computational activities.
10 see http://www.w3.org/TR/2015/NOTE-hcls-dataset-20150514/#s7n_1 section 7.1.2 for details 3
In this paper, we present an ontology pattern named Smart Container that contextualizes computational experiments using Docker as an infrastructure example. We populated our ontology design pattern by analyzing main concepts in computational experiment and aligned with PROV-O, CSO and ACT to provide possibilities for wider extensions.
Acknowledgement We acknowledge funding from NSF grant PHY-1247316 \DASPOS: Data and Software Preservation for Open Science" and the Center for Research Computing at UND. We also acknowledge helpful conversations with Michelle Cheatham, Stian Soiland-Reyes, Matthew Gamble and Carole Goble.