Semantic Copyright Management of Media Fragments
Roberto García, David Castellà and Rosa Gil Universitat de Lleida, Jaume II 69, Lleida, Spain
Keywords: Digital Rights Management, Copyright, Media Fragment, Web Ontology, Semantic Web, Ontology Engineering.
Abstract: The amount of media in the Web poses many scalability issues and among them copyright management. This problem becomes even bigger when not just the copyright of pieces of content has to be considered, but also media fragments. Fragments and the management of their rights, beyond simple access control, are the centrepiece for media reuse. This can become an enormous market where copyright has to be managed through the whole value chain. To attain the required level of scalability, it is necessary to provide highly expressive rights representations that can be connected to media fragments. Ontologies provide enough expressive power and facilitate the implementation of copyright management solutions that can scale in such a scenario. The proposed Copyright Ontology is based on Semantic Web technologies, which facilitate implementations at the Web scale, can reuse existing recommendations for media fragments identifiers and interoperate with existing standards. To illustrate these benefits, the papers presents a use case where the ontology is used to enable copyright reasoning on top of DDEX data, the industry standard for information exchange along media value chains.
1 INTRODUCTION communicate rights so they can be automatically processed in a scalable way. Digitalisation and the transition to a Web full of However, the issues associated with copyright media, where video already amounts more than half management at a Web scale become even more of online consumer traffic1, have introduced new complex when it goes beyond simple access control scalability requirements like bandwidth exigencies, and takes into account also content reuse and the which technology is rapidly evolving to cope with. whole content value chain. In this case, rights However, there are other limiting factors that are not representations need to be more sophisticated so scaling so well, especially those that have been they can capture the full copyright spectrum. traditionally slow moving like copyright. In addition, as reuse is easier when considering As the amount of content made available through just fragments, spatial or temporal, of existing the Web grows, for instance 72 hours of video are content and not full content pieces. Proposed uploaded to YouTube every minute2, the problem of solutions should scale not just to a Web of media but managing its copyright becomes even more relevant. also to a Web of media fragments. Fragments, Consequently, there is already a need to make rights accompanied by scalable copyright management for management scale to a web of media, as pointed by the full value chain, enable a potentially enormous recent initiatives like the PLUS Coalition3 or the re-use market. Linked Content Coalition4. These initiatives, among The main contribution described in this paper is a other things, propose ways to represent and Web ontology for the representation and communication of rights and licensing terms over media assets in terms of their fragments. The ontology is based on Semantic Web technologies 1 Cisco's Visual Networking Index, http://www.cisco.com/en/US/ and integrates with the W3C Media Fragments netsol/ns827/networking_solutions_white_papers_list.html 2 Recommendation (Troncy et al., 2012) to define and YouTube Statistics, http://www.youtube.com/yt/press/statistics. describe spatial and temporal media fragments. html 3 The ontology makes it possible to underpin the PLUS Coalition, http://www.useplus.com media discovery and usage negotiation process, 4Linked Content Coalition, http://www.linkedcontentcoalition.org
230 García R., Castellà D. and Gil R.. Semantic Copyright Management of Media Fragments. DOI: 10.5220/0004493102300237 In Proceedings of the 2nd International Conference on Data Technologies and Applications (DATA-2013), pages 230-237 ISBN: 978-989-8565-67-9 Copyright c 2013 SCITEPRESS (Science and Technology Publications, Lda.) SemanticCopyrightManagementofMediaFragments
facilitating the automation of functionalities for the great efforts in place, the complexity of the rights management. Based on an explicit and copyright domain makes it very difficult to produce interoperable semantic representation for the and maintain implementations based on this communication of rights, the ontology facilitates approach. assessing the reusability of a given media asset The implementers must build them from fragment and eases bringing content onto this specifications that just formalise the grammar of the flourishing market. For instance, by interoperating language and force the interpretation and manual with DDEX data5, one of the main standards for implementation of the underlying semantics. This automating the exchange of information along the has been feasible for less complex domains, for digital supply chain. instance when implementing a MPEG-4 player from The rest of the papers is organised as follows. the corresponding specification. However, this is First, in Section 2, related work is presented together hardly affordable for a more complex and open with the W3C Media Fragments recommendation domain like copyright, which also requires a great that makes it possible to attach licenses to media degree of flexibility. fragments. Then, the Copyright Ontology is Moreover, the limited expressivity of the presented in Section 3 and a use case showing this technical solutions currently employed makes it very ontology in practice is included in Section 4. Finally, difficult to accommodate copyright law into DRM Section 5 presents the conclusions and the future systems. Consequently, DRM standards tend to work. follow the traditional access control approach. They concentrate their efforts in the last copyright value chain step, content consumption, and provide limited 2 RELATED WORK support for the other steps. In fact, just Internet publishing risks are considered and the response is to look for more The DRM Watch review on DRM standards (Rosenblatt, 2008) shows that interoperability is a restrictive and secure mechanism to avoid access control circumvention. This makes DRM even less key issue for DRM systems. For instance, it arises in the content distribution scenario when users want to flexible because it ties implementations to consume content in any of the devices they own. proprietary and closed hardware and software security mechanisms. Interoperability is also critical in the organisation scenario, when content flows through organisations The limited support for copyright law is also a or external content is used in order to derive new concern for users and has been criticised, for instance by the Electronic Frontier Foundation one. The main response to DRM interoperability (Doctorow, 2005). The consequence of this lack is requirements has been the settlement of many basically that DRM systems fail to accommodate rights reserved to the public under national copyright standardisation efforts. The main ones are ISO/IEC MPEG-21 (Wang et al., 2005) and ODRL (Iannella regimes (Springer and García, 2008). 2002), and in both cases the main interoperability Consequently, the DRM world remains apart from the underlying copyright legal framework. As facilitation component is a Rights Expression Language (REL). it has been noted, this is a risk because DRM The REL is a XML Schema that defines the systems might then incur into confusing legal situations. Moreover, it is also a lost opportunity grammar of a license modelling language, so it is based on a syntax formalisation approach. There is because, from our point of view, ignoring copyright law is also ignoring a mechanism to achieve also the MPEG-21 Rights Data Dictionary and a interoperability. Therefore, DRM must evolve to ODRL Data Dictionary Schema (DD) that captures the semantics of the terms employed in the REL, but Copyright Management. It is true that copyright law diverges depending it does so without defining formal semantics (García on local regimes but, as the World Intellectual and Delgado, 2005). 6 This syntax-based approach is also common to Property Organisation promotes, there is a common legal base and fruitful efforts towards a greater level other DRM interoperability efforts and one of main causes of the proliferation of interoperability of copyright law worldwide harmonisation. initiatives that cannot interoperate among them, like A new approach is necessary if we want profit in the e-books domain (Rosenblatt, 2009). Despite
6 WIPO, World Intellectual Property Organization, 5DDEX, http://www.ddex.net http://www.wipo.int
231 DATA2013-2ndInternationalConferenceonDataManagementTechnologiesandApplications
from the Internet as a content sharing medium. The Media analysis techniques can be also used to existence of this opportunity is clear when we create semantic media fragment descriptions, which observe the success of the Creative Commons permit the connection of self-contained media initiative, whose objective is to promote content fragments to the concepts (things, people, locations, sharing and reuse thorough innovative copyright and events...) they are perceived as representing. licensing schemes. Semantic descriptions can be also derived from However, despite the success of Creative existing metadata generated in the media production Commons licenses, this initiative is not seen as an process and augmented by tools provided within the alternative to DRM. The main reason is the lack of media creation phase. flexibility of the available licensing terms. There are Semantic technology is a means to describe mainly six different Creative Commons licenses, all media in a way that can be understood and processed of them non-commercial, and just an informal by machines. Concepts can be unambiguously mechanism for extension and adoption of alternative identified by URIs using Semantic Web Linked Data licensing schemes, CC+7. principles (Hausenblas et al., 2009). Ontologies, Moreover, Creative Commons licenses are which define permitted terms and how they relate to available in three formats: a legal version for one another, are the basis for machine reasoning and lawyers, a more readable version for average users automatic derivation of new knowledge about the and as metadata for computers consumption. media (e.g. a fragment which shows Angela Merkel However, the Creative Commons metadata is not a is also showing the German Chancellor). formal representation of the licenses; it just provides The W3C Media Fragment URI specification a reduced set of terms for building computer- serves as a media format independent, standardised oriented licenses. There are three kinds of means of addressing parts of media resources using permissions (reproduction, distribution and URIs, for instance as shown in Table 1. derivative works), one prohibition (commercial use) The use of the Media Fragment URIs provides a and four requirements (attribution, notice, share consistent identification of fragments in all stages of alike and source code). the media workflow as well as re-use of current tools Consequently, although it is possible to provide and services which support the specification. computer support for simple services like content Moreover, it becomes trivial to enrich media search, there are no mechanisms for customisation fragments descriptions with semantic data based on and advanced computerised support that enable an Semantic Web technologies, which use URIs as the Internet-wide copyright-based alternative to DRM way to identify resources. systems. For instance, the Copyright Ontology, described in the next section, makes use of fragment URIs and 2.1 Media Fragments can attach to them information about their rights situations, licensing terms, etc. Media fragments are defined by spatial or temporal boundaries in media assets. For temporal Table 1: Media Fragment URI example. boundaries, they are based on a start time point and http://my.tv/video.ogv#t=60,100&xywh=12,12,42,30 an end time point (or a duration) that define a Time Spatial fragment, temporal subset of the original media. This kind of fragment, rectangle from pixel fragments can be also defined for audio content. from 60s. x=12, y=12 and width Spatial boundaries are specific to visual media to 100s. 42px, height 30px (pictures or videos) and correspond to a subarea in the original media. The subarea is usually shaped as a rectangle defined by two points or one point plus a 3 THE COPYRIGHT ONTOLOGY height and width. The point coordinates and the sizes are usually defined using pixels or percentages. The Copyright Ontology has been engineered To make media fragments scale, its creation can following the Methontology (Gómez-Pérez et al., be automated based on media analysis techniques 2004) methodology for ontology engineering. It capable of determining appropriate spatial and provides guidance for ontology development process temporal boundaries in visual media, in which a self- but also for other support and management contained part of the media can be found. activities. The ontology developing process it proposes is composed by the following phases: specification, conceptualisation, formalization, 7http://wiki.creativecommons.org/CCPlus
232 SemanticCopyrightManagementofMediaFragments
implementation and maintenance. Table 2 shows this case roles selection, which is The specification phase corresponds to the pre- organised in four classes of generic case roles, which development aspects, where the development are shown at the top, and six categories, which are requirements are identified. The maintenance phase shown at the right. These categories correspond to is a post-development activity, it is performed once verb semantic facets, not disjoint classes of verbs. the ontology is developed. During the Therefore, the same verb concept can present one or conceptualisation activity, the domain knowledge is more of these facets. For instance, the play verb can structured as meaningful models. The static part of show the action, temporal and spatial facets in a the conceptualisation corresponds to the concepts particular sentence. called continuants or endurants (Gangemi, 2002). Then it is time to the dynamic part, which Table 2: Case roles for the NL-oriented pattern. corresponds to the concepts called ocurrents or initiator resource goal essence perdurants (Gangemi, 2002). The process is inspired Action agent, instrument result, patient, by the way we actually model the dynamic aspects effector recipient theme Process agent, matter result, patient, of the world using our main knowledge origin recipient theme representation tools, i.e. natural language. The Transfer agent, instrument, experiencer, theme origin medium recipient central piece is the verb, which models the dynamic Spatial origin path destination location aspects and constitutes the core of natural language Temporal start duration completion pointInTime sentences. Ambient reason manner aim, condition consequence The objective is to apply this same pattern when modelling the dynamic aspects of an ontology. The Consequently, once the verb concepts have been first step is to identify the verb concepts identified, the second step of the proposed pattern corresponding to the ocurrents in the domain at corresponds to the process of determining the case hand, i.e. processes, situations, events, etc. These roles that are necessary to build the dynamic model. concepts will constitute the main part of the model Formal methods can be employed to constraint how for the dynamic part, just the same role verbs play in the verb concept and the case roles are related. NL sentences. Therefore, this pattern allows a great range of model This first step just identifies some concepts that detail levels. Moreover, it is a very complete set of are not enough to build complex knowledge case roles. It includes all the case roles identified in expressions. In order to do that, the inspiration is the refereed bibliography and, as it is shown in the also from how NL sentences work. In NL sentences, next subsection, it has been used during the the verb is connected to other sentence constituents, Copyright Ontology development. During this i.e. participants, in order to build expressions that development process no case role lack was detected model processes, events, situations, etc. This kind of and all the verb models could be built with just the connection has been studied for long in the NL case roles in Table 2. domain and a characterisation of them has been made. These connections are characterised as verb 3.1 Conceptualisation fillers called case roles or thematic roles (McRae et al., 1997). This section details the Copyright Ontology This approach has been extensively used in the conceptualisation activity. This activity is used as an NL research domain but there is little work about illustrative example of the pattern presented in the applying case roles for knowledge representation. previous section, which was employed in the There is the FrameNet (Fillmore et al., 2003) Copyright Ontology engineering process. initiative but it is mainly oriented towards The copyright domain is a complex one and knowledge acquisition from NL sources by semi- conceptualising it is a very challenging task. The automatic annotation. conceptualisation process, as it has been shown in Two of the main proposals about the application the pattern description, is divided into two phases. of case roles for knowledge representation are those The first one concentrates on the static aspects of the for Sowa (Sowa, 2004) and Dick (Dick, 1991). From domain. The static aspects are divided into two these sources, a selection of case roles that can be different submodels due to its complexity. extensively used to model the dynamic part of First, there is the creation submodel. This model ontologies has been built. The contribution of this is the basis for building the conceptual models of the selection is that it is specially tailored to be rest of the parts. It defines the different forms a integrated as a pattern for ontology engineering. creation can take, which are classified following the
233 DATA2013-2ndInternationalConferenceonDataManagementTechnologiesandApplications
three main points of view as proposed by many are broadcast or make available. upper ontologies, e.g. the Suggested Upper Merged Transformation Right: derive. Some Ontology (Niles and Pease 2001): specialisations are adapt or translate. Abstract: Work. At this point we have completed the first phase of Object: Manifestation, Fixation and Instance. the dynamic model part, i.e. the verb concepts have Process: Performance and Communication. been identified. They constitute the key elements in order to build knowledge expressions that represent A part from identifying the key concepts in the the processes, events and situations that occur in the creation submodel, it also includes some relations copyright domain. among them and a set of constraints on how they are In order to build this expression and relate the interrelated. More details for this point and the verb concepts to the other participants, i.e. concepts following steps in the conceptualisation process are in the creation submodel or reused from other available from8. ontologies, it is time to complete the dynamic model Second, there is the rights submodel, which is and detail for each verb concept the corresponding also part of the static part model. The Rights Model case roles. follows the World Intellectual Property Organisation Due to space limitations, this section includes (WIPO9) recommendations in order to define the just the detailed model for the Copy action, which is rights hierarchy. The most relevant rights in the formally known as Reproduce. However, it is DRM context are economic rights as they are related commonly referred to as Copy and this term is the to productive and commercial aspects of copyright. one that is going to be used in the ontology in order All the specific rights in copyright law are modelled to improve its usability. Copies have been as concepts. For the economic aspects of copyright traditionally the basic medium for Work there are the following rights: Reproduction, commercialisation. They are produced from a Distribution, Public Performance, Fixation, Manifestation, from a Fixation of a Performance or Communication and Transformation Right. from another Instance. Therefore, these are the Each right governs a set of actions, i.e. things theme of the Copy verb as it is shown in Table 3. that the actors participating in the copyright life The result is an Instance that is the item cycle can perform on the entities in the creation employed for the physical commercialisation of model. Therefore, it is time to move to the dynamic works, i.e. when a physical item is used as the aspects of the domain. The model for the dynamic vehicle to make the Work arrive to its consumers. part is called the Action Model and it is built on the For example, the making of copies of a protected roots of the two previous ones. work is the act performed by a publisher who wishes Actions correspond to the primitive actions that to distribute copies of a text-based work to the can be performed on the concepts defined in the public, whether in the form of printed copies or creation submodel and which are regulated by the digital media such as CD-ROMs. rights in the rights submodel. For the economic rights, these are the actions: Table 3: Copy case roles. Reproduction Right: reproduce, commonly speaking copy. Case role Range Cardinality Person Distribution Right: distribute. More specifically agent 1..N (Natural or Legal) sell, rent and lend. Manifestation OR Fixation OR theme 1 Public Performance Right: perform; it is Instance regulated by copyright when it is a public result Instance 1 performance and not a private one. pointInTime e.g. ISO8601 1 Fixation Right: fix, or record. location e.g. ISO3166, URL, ... 1 Communication Right: communicate when the ...... subject is an object or retransmit when communicating a performance or previous The central part of Figure 1 shows an example communication, e.g. a re-broadcast. Other related model for expression build using the proposed actions, which depend on the intended audience, pattern as it is applied to the Copy verb concept.
This kind of action patterns are also used to model 8A Semantic Web approach to Digital Rights Management, licenses. Therefore, two additional verb concepts are http://rhizomik.net/~roberto/thesis identified and detailed using case roles: Agree and 9WIPO, http://www.wipo.int
234 SemanticCopyrightManagementofMediaFragments
Disagree. They are the building block of any Language (OWL), concretely on the Description license. Figure 1 shows a license for the Copy action Logic (DL) variant. This implementation can be previously introduced. As it is shown, the condition used to develop a Semantic DRM System based on case role is used in order to introduce a DL reasoning (García and Gil, 2010), as detailed in compensation for the agent that grants the copy the next section. action, a 3€ transfer from the granted agent. As it can be observed in the figure, the condition case role is used to model the obligation deontic 4 USE CASE aspect inherent in copyright licenses. The permission and prohibition deontic aspects also present in The Copyright Ontology has been applied in a real licenses are captured by the Agree and Disagree use case involving media fragments and existing verb concepts and their corresponding theme case DDEX rights data. DDEX data is used in this case as roles. the way to communicate the rights associated to
assets along the value chain. However, DDEX data does just model deals, which capture the kind of actions that can be performed with a particular asset or fragment in a given territory, time point, etc. They do not capture the existing copyright agreements that might make those particular actions legal or not. Table 4 includes a DDEX example.
Table 4: DDEX data example.
The agreement theme corresponds to an implicit
10 Figure 3, so DDEX data can be converted into Copyright Ontology, http://rhizomik.net/ontologies/copyrighton to Semantic Web data based on this ontology. This
235 DATA2013-2ndInternationalConferenceonDataManagementTechnologiesandApplications
Figure 2: Illustration of Copyright Ontology-based reasoning to check if a dispute is supported by existing rights agreements that defines two deals. way, many different deals can be combined and Figure 2. taken into account to decide a dispute. Moreover, This process is based on the instance they can be also combined with other sources of classification service provided by OWL reasoners so information, like existing agreements once they are the implementation effor is reduced to retrieving the also formalised. classes modelling the deals where the intance has been classified into and checking if it is part of an Table 5: DDEX data example modelled using the agreement and thus licensed. It is also checked that Copyright Ontology. the there is no deal the instance has been classified
Once combined, it is possible to use reasoners to 5 CONCLUSIONS AND FUTURE easily implement the process of checking if the WORK dispute being considered is supported by any of the existing deals or agreements. To do that, deals are As the amount of media in the Web increases and modelled as classes based on the intersection or more sophisticated uses like reuse are considered at union of restrictions on the deal action and its case that scale, a way to represent and automatically roles, as shown in Table 5. process media rights becomes even more necessary. These classes define the set of actions that are This problem becomes even more relevant when not authorised by a deal. The reasoner can be then used just the copyright of pieces of content has to be to check if the dispute, modelled as an instance, is inside the set defined by the class and consequently 11Copyright Reasoning Explained, presentation available from it can be interpreted as supported by the deals and MediaMixer Community (free membership). the agreement under consideration, as illustrated in http://community.mediamixer.eu/materials/presentations/copyrigh t/view
236 SemanticCopyrightManagementofMediaFragments
considered, but also media fragments. 2012. Media Fragments URI 1.0 (basic). W3C To provide a scalable solution, we propose using Recommendation, 25 September 2012. highly expressive rights representations that can be http://www.w3.org/TR/media-frags/ connected to media fragments. This proposal is Rosenblatt, B.: 2008 Year in Review: Part 1. DRM Watch, December 28, 2008. http://www.drmwatch.com/drmte materialised into a Copyright Ontology, which is ch/article.php/3793156 based on Semantic Web technologies. The ontology Wang, X., DeMartini, T., Wragg, B., Paramasivam, M., provides a common framework, based on copyright Barlas, C.: The MPEG-21 rights expression language law, capable of giving support across the whole and rights data dictionary. IEEE Transactions on media value chain. Multimedia, 7(3), 408-417, 2005. Existing data formats can be mapped to this Iannella, R.: Open Digital Rights Language (ODRL), common framework and then benefit from formal Version 1.1, 2002. semantics. First of all, media fragment can be García, R., Delgado, J.: An Ontological Approach for the identified using the W3C Media Fragment URI Management of Rights Data Dictionaries. In Moens, M., Spyns, P. (Eds.): Legal Knowledge and recommendation. Moreover, existing data, like Information Systems. IOS Press, Frontiers in Artificial DDEX data used by the industry to communicate Intelligence and Applications, 134, 137-146, 2005. information across the value chain, can also be Rosenblatt, B.: 2009 Year in Review: Part 1. DRM Watch, mapped to the Copyright Ontology. December 28, 2009. http://copyrightandtechnology.co Once integrated and formalised, it becomes m/2009/12/28/2009-year-in-review-part-1/ easier to implement solutions at the Web scale using Doctorow, C.: Critique of NAVSHP (FP6) DRM existing Web ontologies reasoners. This approach Requirements Report. Electronic Frontier Foundation, has been put into practice in a real use case, where 2005. http://www.eff.org/IP/DRM/NAVSHP existing DDEX data is converted into semantic data Springer, M., García, R.: Promoting Music Sampling by Semantic Web-Enhanced DRM Tools. In Grimm, R.; and connected to the Copyright Ontology. Then, Hass, B. (Eds.): Virtual Goods: Technology, Economy, reasoners have been used to help decide if a dispute and Legal Aspects. Nova Science Publishers, 2008. on a media fragment is supported by the existing Hausenblas, M., Troncy, R., Bürger, T., Raimond, Y. DDEX data and copyright agreement and thus it is (2009) Interlinking Multimedia: How to Apply Linked possible to ask for compensation. Data Principles to Multimedia Fragments. LDOW Future work now is to bundle this solution into Workshop 2009, CEUR Workshop Proceedings, 538. an existing asset management system like Fedora Gómez-Pérez, A., Fernández-López, M., Corcho, O.: Commons12, which is also incorporating semantic Ontology Engineering. Advanced Information and technologies and media fragments capabilities. This Knowledge Processing Series. Springer, 2004. Gangemi, A., Guarino, N., Masolo, C., Oltramari, A., setting would help further evaluation the Copyright Schneider, L.: Sweetening Ontologies with DOLCE. Ontology in real use cases, pushing boundaries to In A. Gómez-Pérez, A. & R. Benjamins (Eds.): test its scalability and incorporating other rights data Knowledge Engineering and Knowledge Management, sources, like rights agreements automatically pp. 166-181, Springer, 2002. processed using Natural Language techniques. McRae, K., Ferretti, T. R., Amyote, L.: Thematic Roles as Verb-specific Concepts. Language and Cognitive Processes Journal, 12(2), 137-176, 1997. Fillmore, C. J., Johnson, C. R., Petruck, M. R. L. 2003. ACKNOWLEDGEMENTS Background to FrameNet. International Journal of Lexicography, 16(3), 235-250. The work described in this paper has been partially Sowa, J. F. 2004. Common Logic Controlled English. supported by the European project MediaMixer Draft, 24 February 2004. Available from (Community and Networking for the Remixing of http://www.jfsowa.com/clce/specs.htm Online Media, FP7-ICT-318101) and the Spanish Dick, J.P. 1991. A conceptual, case-relation project InDAGuS (Infrastructures for Sustainable representation of text for intelligent retrieval. PhD Thesis. University of Toronto, Canada. Open Government Data with Geospatial Features, Niles, I., Pease, A. 2001. Towards a Standard Upper TIN2012-37826-C02). Ontology. In C. Welty & B. Smith (Eds.) Proceedings of the 2nd International Conference on Formal Ontology in Information Systems, FOIS (pp. 2-9). REFERENCES Maine, USA. García, R., Gil, R. (2010). Content value chains modelling using a copyright ontology. Information Systems, Troncy, R., Mannens, E., Pfeiffer, S., Van Deursen, D. 35(4), 483–495.
12http://fedora-commons.org
237