The Publishing Workflow Ontology (PWO)

Tracking #: 1301-2513

Authors: 
Aldo Gangemi
Silvio Peroni
David Shotton1
Fabio Vitali

Responsible editor: 
Krzysztof Janowicz

Submission type: 
Ontology Description
Abstract: 
In this paper we introduce the Publishing Workflow Ontology (PWO), i.e., an OWL 2 DL ontology for the description of workflows that is particularly suitable for formalising typical publishing processes such as the publication of articles in journals. We support the presentation with a discussion of all the ontology design patterns that have been reused for modelling the main characteristics of publishing workflows. In addition, we present two possible application of PWO in the publishing and legislative domains.
Full PDF Version: 
Tags: 
Reviewed

Decision/Status: 
Accept

Solicited Reviews:
Click to Expand/Collapse
Review #1
Anonymous submitted on 21/Jan/2016
Suggestion:
Accept
Review Comment:

These are easychair reviews from the WOP special issue that was handled outside of SWJ due to the involvement of multiple SWJ editors. Authors and reviewers have been notified.

Reviewer 1: Boris Villazón-Terrazas

This paper presents an ontology, PWO, for representing information of generic workflows
within publishing processes, e.g., paper publications in journals. The ontology presented in
the paper is part of SPAR Ontologies [1], Semantic Publishing and Referencing Ontologies.
The authors reuses a set of ontology patterns taken from [2], e.g., participation, sequence,
control flow and plan execution, time-indexed situation, and error ontology. Moreover, they
reuse also some vocabularies such as FRBR [3]. Finally, they present three workflow
examples coming from two domains: (1) publishing workflow of a journal paper, and (2)
workflow within the legislative domain.
The paper is well written and organized. The paper compiles the related research efforts
within Workflow modelling.
Review:
Comments
- Section 1 - it would be good to clarify what are the requirements you mention, probably by
using bullets.
- Secttion 2 - is there any related work that covers workflows in publication domain (not
Semantic Web related)
- Section 3 - why did you not reuse other workflow ontologies?, for example the ones
developed in the context of the Workflow4Ever project?
- Section 4 - are there other workflow instances of your ontology, PWO, out there?
- Section 4 - it would be good to provide a small diagram for your second example, the one
from the legislative domain.
- Section 4 - it would be nice to have statistics of the number of classes, properties, etc.

21.08.2015, 16:57
- Section 5 - how did you evaluate the model? how the model cover the set of requirements?
- Section 5 - did you contact Semantic Web Journal editors for using their existing data and
model the upcoming submissions?
[1] http://www.sparontologies.net
[2] http://www.ontologydesignpatterns.org
[3] http://www.ifla.org/frbr-rg

Reviewer 2: Ondrej Zamazal

This paper introduces a pattern-based Publishing Workflow Ontology (PWO). This ontology is
suitable for the description of generic workflows as well as for formalising concrete publishing
processes, e.g. a publication process of article in journal. Authors argue about an importance
of modelling publication processes and provide an overview about existing related workflow-
oriented models approaches. According to workflow definition authors first present a
collection of useful existing ontology design patterns which can be reused for PWO
construction and second they present their application for PWO construction. Finally, authors
demonstrate application of PWO on two different use cases: publishing journal paper in the
Semantic Web Journal and the process of codification of laws of US legislation. These
demonstrations show practical usability of PWO and also its flexibility.
In my opinion, this paper is well-structured, nicely written and it properly exemplifies an
important ontology design principle “reusing of existing ontologies or ontology patterns”.
Although this principle is broadly acceptable, I think that the paper would gain an extra
added-value, if there is some discussion (e.g. extending Section 4) about advantages of
pattern-based approach manifesting traditional benefits of pattern-based approach (and
perhaps also going behind them) for this concrete case of PWO ontology.
Further, is it possible to estimate difficulty of using pattern-based PWO for end-users
modelling their particular workflows such as publishing papers or laws codification versus an
alternative without employing patterns?

Authors state existing related works in Section 2 (e.g. OPMW, P-Plan etc.). I find difficult to
grasp what are the missing features of existing solutions and what are the extra features of
PWO regarding modelling generic or particular workflows. More information about these
points would help to clarify PWO position with regard to existing modelling approches.
As this paper is an extended version of WOP 2014 paper, it would be good to cite the WOP
2014 paper and explicitly state its delta in the introduction.
Ad Table 1 in Section 4.1.1
“produces” PWO entity is described as follows: “The object property linking a workflow step to
anything the step produces, creates or results in: either an activity executing that step, or an
entity
participating in that activity”. Could the result of some workflow step be an activity? I would
assume that the result is an entity, e.g. decision letter or status. I would say that the second
part, “either an activity executing that step, or an entity participating in that activity”, needs
a reformulation.
Ad abstract:
* I think that it is worth mentioning that there are two applications of PWO in the paper.
Minor remark:
* it would be nice if exemplified use cases (with possible modalities) from the paper could be
available online (http://purl.org/spar/pwo). It could serve as further useful documentation
source for users.

Review 3: Raul Palma

The paper describes the Publishing Workflow Ontology (PWO), which enables the
representations of publishing processes. PWO is part of the Semantic Publishing and
Referencing (SPAR) Ontologies suite. The ontology was built mainly by reusing different
ontology patterns, and defining some extensions. In general, the process described for the
creation of the ontology, including the identification of patterns, how they were applied and
extended is relevant and interesting for the special issue. Nevertheless, this contribution
seems more like a typical ontology description rather than a design pattern (the objective of
the special issue) that is a reusable successful solution to a recurrent modelling problem.
Regarding the contribution I have some concern and comments described below.
* Regarding a key modelling decision of separating the workflow (description) and an actual
execution of the workflow, as in the patterns reused and as in other similar ontologies, the
authors did some attempt to address this comment in their workshop paper. They now
introduce class pwo:WorkflowExecution, but it is not really clear how they intend to use this
class; it is introduced as optional to represent the aggregation of steps in a workflow (which
is what pwo:Workflow class also does, or?). This class is not even used in the examples, or
illustrated in Figure 1, and even worse, the pwo ontology file (http://purl.org/spar/pwo) does
not define it. In the examples, it would be clear for me that the workflow description (e.g., of
publishing article in paper) is described with workflow and steps, while an actual execution of
the workflow (i.e., the particular example) is described with workflowExecution and actions
(which execute steps). That is, the paragraph in 4.1 is the workflow description, while the
subsections are describing the workflow execution. Then, it would be easier to reuse the
workflow definition (publishing process) as many times as needed to describe executions,
instead of defining each time the workflow and the execution. If the authors prefer other
modelling choice is fine, but they would need to explain why they don’t follow the same
approach as the patterns they are using.
The ontology is introduced as publishing workflow ontology, however the first statement in
the abstract is that this is an ontology for description of generic workflows, and in order to
show the generality you provide the second example of law codification. This looks, however,
another type publication workflow (publication of a modified/new law), which is fine. My point
is why to present the ontology as publishing workflow ontology and then start by saying that
is a generic workflow ontology. Either you change the name and focus of the paper or just
focus the paper on the publishing workflows (as it is mainly now), and just mention that in
the future you may evaluate its suitability for other types of workflows.

Instead of so verbose examples I would prefer if the paper would include some information
explaining how the ontology would be used for providing the statistics you mention in the
introduction, e.g., SPARQL query with some results. Also I would prefer a discussion on how
this ontology could be actually used in practice, e.g., how you would envision the publisher
system to apply the ontology, and if you see this feasible, i.e., how to get publishers interest.
Do you have some real data ? Do you plan to provide technological support for description of
publishing workflows?
The state of the art discussion is interesting, but I don’t understand what happened between
Section 2 and Section 3. You describe and introduce some approaches, ontologies, and then
in Section 3 you start directly with the definition of your ontology. You don’t explain why
those identified approaches, ontologies were not useful, or why they could not be used as
starting point to extend/adapt it to your needs (if required)?
The ontology file is not in line with the description in the paper. First the workflowExecution
(as described above), but also involvesAction range TimeIndexedSituation, i.e., Action which
is not such type, Step is (as in the previous paper).

Other small comments:
* English still needs to be improved throughout the paper.
* Page 1: “...which describe those scholarly data as RDF statements in the Linked Data”. You
mean in the Linked Open Data Cloud? or what you mean?
* Page 2: “...which includes the specification a particular OWL ontology,” -> of a?
* Page 3: Section 3.3 is a bit confusing (it was more clear in the workshop paper). For
example you have action tasks, but you say tasks are distinct from actions, then you say
actions are classified by tasks. In previous version you used activities instead of action when
you meant something different...
* Page 3: “... Basic plan on its turn reuses...” -> revise
* Page 3: “...These patterns are needed to describe the kinds of steps (the term used here for
tasks) in a workflow and in general in publishing workflows..” -> and in particular?
* Table 1: Its not clear why do you need in the last two rows such mappings to property
chains to show how patterns were used. It not really clear the need of this complexity, and I
don’t know even which property describes is? The description column for these two rows may
need a bit more of explanation, e.g., why a step result can be an activity executing that step