Overview of the MPEG-21 Media Contract Ontology

This paper is focused on the description of Media Content Ontology (MCO), a vocabulary and knowledge model able to represents intellectual property right contracts about audiovisual objects. MCO contracts are RDF documents using such vocabulary, able to define the conditions related to the transfer of the exploitation rights with regulative provisions, usually expressed in terms of what is permitted or prohibited. Such clauses and the related modeling are seen not from the point of view as object of reasoning, therefore the corresponding RDF statements are limited to represent the operative clauses in a machine-readable format.
A mapping between the digital and the narrative form of a contract is foreseen.
Finally a software developed for media content management is illustrated, able to create and manage MCO contracts, as well as CEL (Contract Expression Language) contracts. Finally, the integration of MCO and CEL modules is presented.

The paper describes in an organic fashion the work carried out by the MPEG-21 standard working group and in particular the related semantic approach. However the unavailability of such ontology makes it difficult to provide a detailed analysis of it. For example it would be interesting to know at which OWL language profile the presented ontology belongs to, similarly a details documentation of classes and properties would be useful.

As for the deontic notions, the authors seem not up-to-date to the state-of-the-art (the latest references of papers from the AI and Law community are dated back to 2007). In this respect I suggest to have a look at the literature of the latest ICAIL conferences and at AI and Law Journal articles related to the modeling of provisions, deontic notions and their relations.

The authors have also developed a set of tools to editing and manage MCO contracts, among the others in particular a tool to convert between the Contract Expression Language (CEL) contracts and the MCO contracts. Since the two formats depends on XML and RDF/OWL schemas respectively, the authors are invited to comment on the different expressivity power of the approaches. Is there any loss of information in the transformation?

One of the most interesting aspects of the paper, in my view, is the claimed ability of the model to describe conditional statements. This claim however is not well detailed, while a better description of how this is achieved in MCO, as well as a set of example to demonstrate such ability are necessary.

Minor remarks

pag. 3 compactnesas —> compactness
pag. 4 ammendments —> amendments
pag. 4 itslef —> itself
pag. 7 permitting all the its specializations —> permitting all its specializations

The paper describes the MPEG-21 Media Content Ontology, which is an ISO standard for describing multimedia access rights contracts. The paper is nicely written, easy to read and provides both an overview of the motivation behind the standard, its content, and several usage examples, including usage examples of the accompanying software for managing contracts expressed using the ontology.

It should be noted however that the paper does not contain any novel research contributions, as far as I can tell. It is simply and overview of a standard and it accompanying tools. Hence, from a research perspective this paper does not contribute to the field. On the other hand, it helps to disseminate and explain an important standard, and exemplifies a general use case for ontologies, i.e. as reusable standards specifications. For the readers of a special issue on Semantic Web for the legal domain, this could be highly interesting and is certainly relevant. Also, the authors do not try to frame the paper as a research paper, but are very explicit with the fact that it is an overview (e.g. by mentioning it already in the title and abstract).

Reading the paper under the above premises I have very little to comment on the actual content. One question that the authors do not comment on is the issue of modularity - is the ontology modular? Depending on how big the ontology is, it may be interesting to view it as consisting of several module, which can be reused for different purposes. Modularity can also help an casual user to understand the ontology better, than if it is a large monolithic structure.

An additional question is raised when the authors state that it is an OWL2 ontology. What features of OWL2 are actually used in the ontology? Does this simply mean it is an OWL ontology, or that it actually uses some of the features that were added in OWL2 compared to the earlier version of OWL? Overall, since this is a paper submitted for the Semantic Web journal, it would be interesting if the authors could explain a bit more of the actual design of the ontology, what kinds of axioms it contains, and why, etc. Additionally, there are not really any trade-offs mentioned, or lessons learned from the standards development process. Such experiences could also be valuable for the research community, and since the paper is written by the contributors of the standard they should be well aware of all the issues that were encountered in the development process.

One general issue throughout the paper is that the figures are not that clear. What do the arrows and boxes actually mean? What is the notation used? For instance, in Figures 1-7 where the ontology is described, the authors mix between describing axioms, such as FactComposition rdfs:subClassOf mvco:Fact (Figure 5) and something else, e.g. FactComposition hasFact mvco:Fact. The latter may also indicate the presence of an axiom, such as a class restriction on the FactComposition class, but it could also indicate domain an range restrictions on the property - but this is never explained. Similarly, the | character and the alternative names of the classes in Figure 5 are not explained either. Additionally, in Figure 7 the authors use the same notation as for classes to express individuals of those classes. This is quite confusing, and it gets even more confusing from Figure 8 and onwards, where different colours of the boxes and ovals are introduced without explanation, and = signs start to appear inside the boxes. The authors either need to use some well-known visualisation strategy, e.g. as used by some common tool, or they need to carefully explain to the reader what the figures mean, and what notation they use. Also, Figure 9 is strangely broken up into pieces, so that arrows seem to end suddenly, not leading anywhere.

Additional minor issues include:
- Section 3, 2nd paragraph: compactnesas -> compactness
- Section 3, last paragraph: I do not understand what "dig" means in this context
- There are some layout issues throughout the paper, e.g. a portion of page 4 is empty and section 4.2 is broken up in the middle of a paragraph although there is plenty of space for it to continue on page 4.
- In section 6.1.3, step 5 of the CEL to MCO transformation: is it an ontology that is returned, or data (individuals) expressed according to the ontology?
- Section 6.2 last sentence: editing - edited
- Section 7, first sentence: in -> on

This paper is a description of the MPEG-21 Media Contract Ontology (MCO), a recently approved standard. The authors of the paper are also the editors of the ontology. The paper is well written, clearly structured, and falls into the scope of this journal and this particular special issue. The authors, however, have classified the paper as a 'survey article’, and as such it is not comprehensive and balanced enough, and consequently not suited as an introductory text for researchers, PhD students of practitioners. In my view, it is a rather a ‘description of ontology’, since it focusses mainly on the MPEG-21 Media Contract Ontology, and, although the work described in the paper is well situated in relationship to similar efforts, related approaches are only mentioned briefly.

As a ‘description of ontology’, the principles and methodologies that guided the design of the ontology are, in my opinion, not sufficiently explained. A pointer to a first set of requirements is provided (ref. [7]), but not further discussed in the paper. A set of criteria is mentioned (making reference to an experiment, see [8]) and briefly listed, but without going further into the details.

Consequently, what I am missing in this paper is a clear justification of the expressive power required for MCO. MCO is formalised in OWL2, but then, after a look at the mco-core.owl and mco-ipre.owl files available online, only a very limited part of the expressive power of OWL2 is actually used. Therefore, it is not clear to me what the gain of the current status of MCO is with respect to a less-expressive representation language (e.g., the XML-based language CEL).

As objective of MCO, the authors mention machine-readability to achieve better metadata interoperability and systems integration. But this is too vague an objective, and more detail is necessary to justify the choice of representation language and the granularity of detail specified in the ontology. For example, I would like to see a more detailed discussion of the sort of reasoning services that are expected to be provided by using MCO.

There are a couple of design decisions of MCO that are still unclear to me and should be clarified in the final version of the article:

(1) As far as I see, in MCO every instance of mvco:Fact is explicitly related to its truth value via the mvco:isTrue datatype property. But by introducing Boolean operators into MCO to represent more complex conditions of deontic expressions by means of non-atomic propositions, how is the consistency of truth values between the composite proposition and its atomic propositions maintained?

(2) Although several of the classes of the MCO core ontology (and its extension for the exploitation of IPR) are defined, in many parts these ontologies are just taxonomies of terms. Consequently, it is hard to grasp the reasoning power one could get out of them. For example, there are no disjointness axioms for any of the subclasses of mco-ipre:ExploitationCondition (they could all be interpreted as the same set), and there are no logical relationships stated between Linear and NonLinear, FreeOfCharge and Pay, PayPerView and Subscription, Open and Restricted, Limited and Unlimited, etc. What sort of reasoning is envisioned? Because as it stands it is not possible to distinguish between different conditions.

In the conclusions the author claim that the different examples provided “confirm the usefulness of MCO and how to use it”. In my opinion, only by showing how contracts would be represented in MCO is not enough to assess the usefulness of the ontology. It is by demonstrating in detail the reasoning that can be done with the ontology when the usefulness can be assessed, i.e., once the required reasoning services are clearly specified. This sort of evaluation is still missing.

I also was surprised that in Sections 6.1.2 and 6.1.3 on CEL and the MCO-CEL mapping there is no detailed discussion on the concrete loss of information occurring when mapping from MCO to CEL since their expressiveness cannot be equivalent.

Finally, some minor typos I have noticed while reading the paper:

page 2: “a determined criteria” => “a determined criterion” or “determined criteria”
page 6: Table 1 misses the triple “:Contract rdf:type mco-core:Constract”
page 7: “all the its” => “all its”
page 9: in Table 2 the first two triples should be “:user001 rdf:type mvco:User” and “:user002 rdf:type mvco:User”
page 19: “uniquely identifier” => “unique identifier”
page 20: “sub-clause” => did you mean “sub-section”?
page 21: the reference to Fig 17 is repetitive

Finally the are several references to sections and figures which are erroneous:
- page 2: all of those in last paragraph of Section 1
- page 14: Figures 9 and 10 => should be 10 and 11
- page 18: reference to Section 5.2 should be 6.2
- page 19: reference to Section 5.1.3 should be 6.1.3

This is a remark by the Editor-in-chief. This paper was withdrawn by the authors before the review process was complete. It seems there was some confusion regarding the paper type and review criteria, therefore the withdrawal. We may receive a fresh submission under a different paper type.