This document is licensed under a Creative Commons Attribution 3.0 License.
This document is merely a public working draft of a potential specification. It has no official standing of any kind and does not represent the support or consensus of any standards organisation.
Scientific workflows are used to describe series of structured activities and computations that arise in scientific problem-solving, providing scientists from virtually any discipline with a means to specify and enact their experiments. From a computational perspective, such experiments (workflows) can be defined as directed acyclic graphs where the nodes correspond to analysis operations, which can be supplied locally or by third party web services, and where the edges specify the flow of data between those operations.
Besides being useful to describe and execute computations, workflows also allow encoding of scientific methods and know-how. Hence they are valuable objects from a scholarly point of view, for several reasons: (i) to allow assessment of the reproducability of results; (ii) to be reused by the same or by a different scientist; (iii) to be repurposed for other goals than those for which it was originally built; (iv) to validate the method that led to a new scientific insight; (v) to serve as live-tutorials, exposing how to take advantage of existing data infrastructure, etc. This follows a trend that can be observed in disciplines such as Biology and Astronomy, with other types of objects, such as databases, increasingly becoming part of the research outcomes of an individual or a group, and hence also being shared, cited, reused, versioned, etc.
Challenges for the preservation of scientific workflows in data intensive science, include: (a) the consideration of complex digital objects that comprise both their static and dynamic aspects, including workflow models, the provenance of their executions, and interconnections between workflows and related resources, (b) the provision of access, manipulation, sharing, reuse and evolution functions to these complex digital objects, (c) integral lifecycle management functions for workflows and their associated materials.
Thus the use of workflow specifications on their own does not guarantee to support reusability, shareability, reproducibility, or better understanding of scientific methods. Workflow environment tools evolve across the years, or they may even disappear. The services and tools used by the workflow may change or evolve too. Finally, the data used by the workflow may be updated or no longer available. To overcome these issues, additional information may be needed. This includes annotations to describe the operations performed by the workflow; annotations to provide details like authors, versions, citations, etc.; links to other resources, such as the provenance of the results obtained by executing the workflow, datasets used as input, etc.. Such additional annotations enable a comprehensive view of the experiment, and encourage inspection of the different elements of that experiment, providing the scientist with a picture of the strengths and weaknesses of the digital experiment in relation to decay, adaptability, stability, etc.
These richly annotation objects are what we call workflow-centric Research Objects. The notion of Research Object (as discussed in [BECHHOFER11]) is a general idea that aims to extend traditional publication mechanisms and take us "beyond the pdf" [FORCE11]]. An RO is an aggregation of resources along with annotations on those resources. The aggregation itself may also be annotated, where by annotation, we mean the association of arbitrary additional information with a resource. The Research Object thus collects together relevant resources along with annotations that enable the understanding, reuse etc. of its constituent parts. In a workflow-centric Reearch Object describing an investigation for example, annotations could describe how data sources have been used or how intermediate results were derived. Executable papers [EXEC] similarly aim to support validation, exploration and interaction in publication in order to support validation. Hunter [HUNTER06] proposes the notion of Scientific Publication Packages (SPP) to describe "the selective encapsulation of raw data, derived products, algorithms, software and textual publications". SPPs are motivated primarily by the need to create archives for the variety of artifacts produced during the course of a scientific investigation; they ideally contain data, methods, software and documents, but also their provenance as well.
The focus of Wf4Ever is on workflow preservation (and more specifically workflows supporting scientific investigation). Thus the objects that will be described using the model are inherently workflow-centric. Although the basic infrastructure (aggregation + annotation + domain vocabularies) that the RO model supports is applicable to many situations (for example the executable or enhanced publication of [FORCE11]), the specific vocabularies defined within the Wf4Ever RO model are intended to support those Research Objects that have workflows (methods) as their primary content.
The intention is that a Wf4Ever Research Object will aggregate information about a workflow (or collection of workflows) possibly including details of its execution trace, the data items consumed or produced by the workflow, plus provenance information about the lineage of the workflow, data items or the aggregation itself. This intention is that this information will support reproducability of the workflows, reuse of the components and, perhaps most importantly will facilitate subsequent understanding of the investigation that the workflow is intended to support. The support of "full-fat" reproducability in terms of re-execution of a process in order to produce identical results is problematic, particularly when operating in a distributed, web-service based ecosystem.
This particular document describes the vocabularies that describe Workflow-centric Research Objects within Wf4ever. A complete, functional RO based ecosystem will require a number of different services for creation, storage, manipulation, recommendation, visualisation etc. of Research Objects. These services are not considered here. Nor do we discuss vocabularies or mechanisms that support Research Object evolution.
This document describes the v0.1 release of the ontologies.
Research Objects play multiple roles. In the first case, they are technical objects. They provide access to the resources that are needed to support execution of investigations and record the provenance traces of those executions. They encapsulate dependencies between resources and maintain versioning information about the lineage and evolution of those resources.
At the same time, they are social objects. They encapsulate reusable protocols and know-how. They record best practices and support reproducability. They are citable artifacts that can be referred to and quoted. They record and represent information about the people involved in investigations -- those who create, use, extend and curate the objects.
These roles bring requirements on the representation structure and vocabularies used to describe Research Objects. This specification focuses on the technical aspects and describes the core Wf4Ever Research Object vocabularies that provide container structures and vocabulary for describing workflow objects. Additional vocabularies covering evolution, lifecycle, versioning and other social aspects will be covered elsewhere.
The Wf4Ever Research Object model consists of a suite of ontologies or vocabularies that are used to describe workflow-centric ROs. The key ontologies provided are:
Provides basic structure for the description of aggregated resources and the annotations that are made on those resources.
A vocabulary for the description of workflows. This provides an abstraction that can be mapped to different particular workflow systems.
A vocabulary for the description of provenance information. This provides an abstraction that can be mapped to different provenance vocabularies.
The RO Model uses two existing vocabularies to provide aggregation and annotation functionality. Object Reuse and Exchange [ORE] is used for specifying aggregation of resources and the Annotation Ontology [AO] is used to represent annotations.
Each of these ontologies are described in more detail below.
A research object aggregates a number of resources that are used and/or produced in a given scientific investigation. This aggregation supports access to those collected resources – or at least access to the identification of those resources. The resources aggregated in an RO can be distributed, thus a particular user may be allowed to see the metadata associated with a resource without necessarily having access to its content. For example, dereferencing a URI which is aggregated in a Research Object may require further authentication. A research object allows for annotation of the resources aggregated in it, where by annotation we mean the general notion of decorating a resource with some additional, arbitrary information (@@NOTE: the usage of the word "aggregated" to describe the collecting of resources in the RO rather than "contained"). The aggregation provides a container within which these annotations can be asserted, allowing for the capture of context and provenance concerning those relations and annotations. Such context is needed to support, for example, measures of trust of quality.
The aggregated resources of an RO could include workflow descriptions, workflow runs, artifacts, etc. The figure below illustrates a research object described using a manifest. It also illustrates the specific class of research object named workflow research object, which refers to research objects that contain at least one workflow description.
The research object is described in the ro
ontology under the namespace
This ontology extends the OAI-ORE ontology.
has characteristics: functional
The wfdesc ontology describes an abstract workflow description structure, which on the top level is defined as a wfdesc:Workflow.
A wfdesc:Workflow contains several wfdesc:Process instances, associated using the wfdesc:hasSubProcess property. Each of these (and the workflow itself) wfdesc:hasInput and wfdesc:hasOutput some wfdesc:Parameter (wfdesc:Input or wfdesc:Output). An wfdesc:Artifact is associated with a wfdesc:Parameter using wfdesc:hasArtifact. The wfdesc:Workflow also wfdesc:hasDataLink several wfdesc:DataLink instances, which forms the connection between parameters. Thus this ontology allows the description a direct acyclic graph, or a dataflow.
This ontology is meant as an upper ontology for more specific workflow definitions, and as a way to express abstract workflows, which could either be hand-crafted by users ("ideal workflow description") or extracted from workflow definitions of existing workflow systems, like Taverna's .t2flow and Scufl2 formats.
The wfprov ontology shows how to link these workflow descriptions to a provenance trace of a workflow execution.
The example below illustrates a workflow wf1 that is composed of three processors :procA, :wf2 and :procC, which are connected in sequence using data links. :wf2 is a subworkflow that contains the process :procB.
@prefix wfdesc: <http://purl.org/wf4ever/wfdesc#> . :wf1 a wfdesc:WorkflowTemplate ; wfdesc:hasSubWorkflow :wf2 ; wfdesc:hasSubProcess :procA, :procC . :procA a wfdesc:Process ; wfdesc:hasOutput :param1 . :procC a wfdesc:Process ; wfdesc:hasInput :param2 ; wfdesc:hasOutput :param3 . :wf2 a wfdesc:WorkflowTemplate ; wfdesc:hasInput :param4 ; wfdesc:hasOutput :param5 ; wfdesc:hasProcess :procB . :procB a wfdesc:Process ; wfdesc:hasInput :param6 ; wfdesc:hasOutput :param7 . :innerWorkflow wfdesc:hasDataLink [ wfdesc:hasSource :param4; wfdesc:hasSink :param6 ], [ wfdesc:hasSource :param7; wfdesc:hasSink :param5 ] . :wf1 wfdesc:hasDataLink [ wfdesc:hasSource :param1; wfdesc:hasSink :param4 ], [ wfdesc:hasSource :param5; wfdesc:hasSink :param2 ] .
The wfprov ontology describes the provenance of the workflow results. A wfprov:WorkflowRun describes the activity of running a wfdesc:WorkflowInstance. It groups the wfprov:ProcessRuns of the execution with the relationship wfprov:wasPartOfWorkflowRun.
Each wfprov:ProcessRun represents a step of the workflow execution. It may have dependencies to different wfprov:Artifacts, which are modeled with the relationship wfprov:usedInput. Every result generated by one of the wfprov:ProcessRuns is linked by the relation wfprov:wasOutputFrom
Different wfprov:ProcessRuns may have been executed by different workflow engines. We can specify which one run each step with the wfprov:wasEnactedBy relationship.
Finally, we can link each wfprov:Artifact, wfprov:ProcessRun and wfprov:WorkflowRun to their correspondant wfdesc description by the wfprov:describedByParameter, wfprov:describedByProcess and wfprov:describedByWorkflow respectively.
The example below illustrates a workflow execution we1 that was composed of two processes :proc1 and :proc2. The output of :proc1 (:o1) is the input used by :proc2. Both processes are described by a wfdesc:Process
@prefix wfprov: <http://purl.org/wf4ever/wfprov#> . :wf1 a wfprov:WorkflowRun ; wfprov:describedByWorkflow :wfTempl; wfprov:wasEnactedBy :Taverna. #Additional metadata about the workflow run could be added here: start time, end time, whether it has bees successful, etc. :proc1 a wfprov:ProcessRun ; wfprov:usedInput :i1 ; wfprov:wasPartOfWorkflowRun :wf1; wfprov:describedByProcess :templProcess1. :proc2 a wfprov:ProcessRun ; wfprov:usedIntput :o1 ; wfprov:wasPartOfWorkflowRun :wf1; wfprov:describedByProcess :templProcess2 . :i1 a wfprov:Artifact; wfprov:describedByParameter :param1. :o1 a wfprov:Artifact; wfprov:wasOutputFrom :proc1; wfprov:describedByParameter :param2. :o2 a wfprov:Artifact; wfprov:wasOutputFrom :proc2; wfprov:describedByparameter :param3. #wfdesc:parameters and templates have been omitted to leave the example as simple as possible.
The wf4ever ontology specifies extensions that specific to Wf4Ever, and probably not as liable for reuse as the previous ontologies. This ontology imports all the other ontologies described on this page, and so can also be used to get a complete picture of all the Wf4Ever ontologies. The namespace is http://purl.org/wf4ever/wf4ever#
Here is an example research object containing a single workflow a_workflow.t2flow and an annotation which as been added by a software agent.
Here we can see that Stian has made a research object at 15:01:10, and at 15:02:10 he added <a_workflow.t2flow> to the research object. Note that in this case we do not have informatio about when <a_workflow.t2flow> was created or by whom. The research object also aggregates an annotation :ann1. This annotation was created by "t2flow workflow annotation extractor" a few seconds after Stian added <a_workflow.t2flow> and is an annotation on the aggregated resource. We are also told that the content of annotation <.ro/ann1> was created by Marco more than a year earlier. The original annotation by Marco was found inside the t2flow by the extractor.
This example serves to illustrate that annotations can be made on both the entire RO and resources aggregated within it.
Resolving the annotation body gives:
Inside the annotation body, which we have been told was done by Marco, we find an abstract version of the .t2flow workflow (using the wfdesc ontology) which has been extracted by the agent. This reveals some workflow annotations done by Marco in 2010; title and description on the workflow, and descriptions for the input and output ports.
This example comes from a use case that deals with large sets of tabular data (catalogues of numbers) curated by the user. The data are calculated by means of mathematical equations based on experimental values coming from external data repositories. Updating of the external data repositories has an impact on the curation of the user dataset, it is essential to know how (in which quantity) and when these databases are updated, so that the propagation of these changes through the existing internal relation among the data can be triggered and registered.The scientific experiment represented by this research object pertains to the multi-wavelength study for a sample of the most isolated galaxies in the local universe. This study characterizes each galaxy of this sample through both the measurement of basic astrophysical properties:
Specifically, this research object is focused on the calculation of the intrinsic luminosity in the Johnson B-band, in order to achieve it the measurement or calculation of all those astrophysical properties is needed.
This research object consists of several resources, including workflows, input and output datasets, scripts, web services and other documents. The type and role of each of these resources (by their relative name) are as follows:
Here we provide snippets of RDF illustrating particular aspects of the RO description. These snippets are all taken from the RO manifest.
<http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/manifest> dct:created "2012-02-01T09:33:56.923Z"^^<http://www.w3.org/2001/XMLSchema#dateTime> ; dct:creator [ a <http://xmlns.com/foaf/0.1/Agent> ; foaf:name "RODL" ] ; ore:describes <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/> ; a <http://purl.org/wf4ever/ro#Manifest> .
The following snippet captures the basic aggregation of resources within the RO.
<http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/> dct:created "2012-02-01T09:33:56.832Z"^^<http://www.w3.org/2001/XMLSchema#dateTime> ; dct:creator <https://www.google.com/accounts/o8/id?id=AItOawl7tEZAtdiQmIigvXcU4KkhGsBzgm0gfKM> ; ore:isDescribedBy <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/manifest> ; a <http://purl.org/wf4ever/ro#ResearchObject> ; ore:aggregates <http://leda.univ-lyon1.fr/ledacat.cgi>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/manifest#0f184259-a189-4d2c-8c04-66d7d322d1a8>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/manifest#14e42518-b337-2312-a76b-aa2b21cb3222>, ... <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/manifest#annGathering>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/manifest#annPropagation>, ... <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/CONTENT.txt>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/D1.2.pdf>, ... <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/GoldenTrace.txt>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/NamesLEDA.txt>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/Paturel.pdf>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/README.txt>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/RECIPES.txt>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/Ruiz.pdf>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/Vauglin.pdf>, ... <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/btcNew.txt>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/calculateDistance.py>, ... <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/comparison_and_update_values_475535.>, ... <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/eq2gal.py>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/extractJ2000Coord.py>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/extractPropertyVal.py>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/morphoNew.txt>, .. <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/updating.py>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/velocitiesNew.txt> .
An annotation expressing creation metadata. Jose Enrique Luiz created resources GoldenTrace.txt.
<http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/manifest#14e42518-b337-2312-a76b-aa2b21cb3222> ao:body <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/GoldenTrace.txt> ; dct:created "2012-02-01T09:37:23.662Z"^^<http://www.w3.org/2001/XMLSchema#dateTime> ; dct:creator _:A0 ; ao:annotatesResource <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/> ; a ao:Annotation . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/manifest#14e42518-b337-4698-a76b-aa2b21cb3222> ao:body <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/RECIPES.txt-8230278887054687059> ; dct:created "2012-02-01T09:37:23.662Z"^^<http://www.w3.org/2001/XMLSchema#dateTime> ; dct:creator _:A0 ; ao:annotatesResource <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/RECIPES.txt> ; a ao:Annotation . _:A0 a <http://xmlns.com/foaf/0.1/Agent> ; foaf:name "Jose Enrique Ruiz" .
Finally, we have annotations that describe the fact that a particular workflow run made use of data sets to produce some results.
<http://leda.univ-lyon1.fr/ledacat.cgi> a <http://purl.org/wf4ever/wf4ever#WebServiceProcessTemplate> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/annGathering#GatheringRun1> wfprov:describedByWorkflow <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/gathering_galaxy_properties_using_hyperleda_129473.> ; wfprov:usedInput <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/NamesLEDA.txt> ; a <http://purl.org/wf4ever/wfprov#WorkflowRun> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/NamesLEDA.txt> a <http://purl.org/wf4ever/wf4ever#Dataset> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/agNew.txt> wfprov:wasOutputFrom <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/annGathering#GatheringRun1> ; a <http://purl.org/wf4ever/wf4ever#Dataset> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/btNew.txt> wfprov:wasOutputFrom <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/annGathering#GatheringRun1> ; a <http://purl.org/wf4ever/wf4ever#Dataset> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/extractJ2000Coord.py> a <http://purl.org/wf4ever/wfdesc#Process> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/extractPropertyVal.py> a <http://purl.org/wf4ever/wfdesc#Process> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/gathering_galaxy_properties_using_hyperleda_129473.> wfdesc:hasSubProcess <http://leda.univ-lyon1.fr/ledacat.cgi>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/extractJ2000Coord.py>, <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/extractPropertyVal.py> ; a <http://purl.org/wf4ever/wfdesc#Workflow> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/j2000Coords.txt> wfprov:wasOutputFrom <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/annGathering#GatheringRun1> ; a <http://purl.org/wf4ever/wf4ever#Dataset> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/logr25New.txt> wfprov:wasOutputFrom <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/annGathering#GatheringRun1> ; a <http://purl.org/wf4ever/wf4ever#Dataset> . <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/velocitiesNew.txt> wfprov:wasOutputFrom <http://sandbox.wf4ever-project.org/rosrs5/ROs/HyperLEDALuminosities/.ro/annGathering#GatheringRun1> ; a <http://purl.org/wf4ever/wf4ever#Dataset> .
The v0.1 RO ontologies can be found on github as OWL/Turtle format:
Note on namespaces:
No normative references.