Development Issues on Linked Data Weblog
Enrichment
Ivan Ruiz-Rube1, Carlos M. Cornejo1, Juan Manuel Dodero1, and Vicente M.
Garca2
1 Department of Computer Languages and Systems
University of Cadiz
{ivan.ruiz,carlos.cornejo,juanma.dodero}@uca.es
2 Free Software Research and Development Foundation
vmgarcia@fidesol.org
Abstract. In this paper, we describe the issues found during the devel-
opment of LinkedBlog, a Linked Data extension for WordPress blogs. This
extension enables to enrich text-based and video information contained
in blog entries with RDF triples that are suitable to be stored, managed
and exploited by other web-based applications. The issues have to do
with the generality, usability, tracking, depth, security, trustiness and
performance of the linked data enrichment process. The presented an-
notation approach aims at maintaining web-based contents independent
from the underlying ontological model, by providing a loosely coupled
RDFa-based approach in the linked data application. Finally, we detail
how the performance of annotations can be improved through a semantic
reasoner.
Keywords: Linked Data, annotation, RDFa, web blogging
1 Introduction
The Semantic Web promised a new model in which machines were to be able to
understand and process web-based information [3], so that intelligent agents can
nd, combine and act on the information needed by people in their everyday use
of the Web. The Semantic Web hype, however, has gone
at, and today is often
replaced by the term Linked Data or Web of Data, which aims at a less ambitious
goal that has to do with the automated integration of information. Linked Data
refers to the fact of describing and publishing the information of web pages
in a structured manner that can be easily processed by software programs [4].
Semantic web techniques can be still used to integrate information from dierent
sources and obtain value-added features.
Linked Data presents a distinctive feature, i.e. its implementation is some-
what reversed to the traditional data modeling paradigm. While the classical
development approach is rst designing the data model and then populating it
with data instances, the Linked Data approach is to provide structured data

2 Ruiz-Rube et al.
access after the publication of data instances. Providing a structure means en-
riching web pages with annotations, which are usually compliant with W3C
standards such as OWL and RDF(S).
The ease of enriching data held in the web to expose their underlying con-
cepts, structure or schema depends on how the data sources are designed. For
instance, there is a common ground between relational databases and ontology
models. These structured models make easier to dene RDF/OWL classes that
hold the same base properties of a table in a relational database. Nevertheless,
data on the Web are often published as regular text, which is more dicult to
annotate automatically to nd a correspondence with a given ontology.
The issue of adding formal semantics to web contents as metadata is a major
challenge for the Semantic Web community [16]. On one hand, semantic annota-
tions are often based upon the knowledge of the actors about a specic domain
while such annotations are being provided. On the other hand, Natural Language
Processing (NLP) tools can provide annotation on text-based web sources, but
do not have yet the required maturity to overcome this issue. Furthermore, it
is increasingly common to nd multimedia resources such as images and videos
embedded in actual web applications. These resources implicitly contain much
information, so it would also be interesting to describe and annotate the infor-
mation they hold in a similar manner to the annotation of text. In either case,
users can produce or consume web resources and, at the same time, can play the
role of an annotator, publisher or reviewer of the contents [2].
We have built a tool for text and video annotation of blog messages based
on existing, standard ontologies. The blog contents, properly annotated, can
be then exploited for various purposes, such as indexing by a semantic search
engine, or generating tag cloud concept representations. The Linked Data an-
notation approach aims at two objectives: (i) to keep independence from the
web contents or applications that must be enriched; and (ii) to keep indepen-
dence from the underlying ontology models used to describe annotations. These
objectives deal with integration issues described elsewhere [7]. However, in the
fulllment of such goals, other development issues have emerged, such as: multi-
ple annotation of the same concept by several users (i.e. concurrency); the need
of tracking who provides each annotation (i.e. provenance); the eort required
to achieve an acceptable degree of usability and performance of the annotation
tool; and the exploitation of annotations by means of Semantic Web techniques
in order to make users' tasks easier. This paper describes these and other issues
and illustrates them through the implementation of LinkedBlog, a Linked Data
extension to WordPress3 that enables to enrich text and video contents provided
by the blog users.
The rest of the paper is structured as follows. Section 2 describes other works
related to linked data enrichment. In section 3 we deal with the issues found in
the development of the annotation approach. Section 4 presents our approach
for text and video annotation and nally section 5 draws some conclusions and
provides an outlook to future works.
3 http://wordpress.org/

Development Issues on Linked Data Weblog Enrichment 3
2 Related work
Web-based resources must be annotated to convert them into Linked Data re-
sources in order to be machine-understandable. Such Linked Data resources are
held in semantic content management systems that often have the functional-
ity and look-and-feel of wikis |such as DBPedia [5], Freebase [6]| or blogs
|such as Zemanta [21]. These systems provide semantic Application Program-
ming Interfaces (API) [8] in order to host and exploit linked data. On one hand,
Freebase and DBPedia can extract information from the Wikipedia and pub-
lish it as RDF(S), according to their own schemas. Freebase enables to create
special-purpose Linked Data applications (e.g. Thinkbase [12]) on its own host-
ing platform, called ACRE. Freebase can pull out Wikipedia data as well as
DBPedia linked data. If there is the need of linking resource annotations to an
standard ontology, the linked data schemas require an adaptation or transfor-
mation step [18]. This can be seen in Freebase's RDF output, in which a number
of OWL:SameAs RDF identiers point to DBPedia. The Zemanta blogging as-
sistant uses the English Wikipedia to capture knowledge of dierent areas of
interest (e.g books, politics, sports, etc) and enrich web resources with in-text
hyperlinks. At the core of the system there is a collection of semantic entities
that represent concepts. However, they are not endorsed by any standard RDF
schema or ontology.
Instead of having the annotations centralized on a hosting platform, an ap-
proach to distributed annotation has been provided by Annotea [19]. Annotea is
a Web-based shared annotation system based on a general-purpose open RDF
infrastructure. Annotations are external to the documents and can be stored into
special annotation servers [14]. It also provides several plugins for the client side
(e.g. Annozilla4 or Amaya5). Related to externally stored annotations, Google
SideWiki6 is a browser extension that allows web site viewers to add comments
about web contents. The relevancy of these user-generated comments is based
on Google's ranking algorithms and in consequence they are not held upon the
basis of an ontological model.
Multimedia annotation approaches focus on acquiring metadata descriptions
in order to facilitate indexing, search and retrieval. Feng et al. [9] have intro-
duced an automatic image and video annotation technique for retrieval, based
on textual queries, where the images that form the sequence are partitioned into
regions. Their motivation was to improve the expensive task of manual annota-
tions carried out by users (e.g. librarians) by proposing a statistical generative
model that uses a set of annotated training images. These kinds of modeling
outperform reasonable skills in closed domain collections, but the quality of the
approach critically depends on the training set used. Regarding the existence
of similar collaborative applications, a software tool has been developed for the
semantic annotation of video using multimedia ontologies [20]. It is based on an
4 http://annozilla.mozdev.org/
5 http://www.w3.org/Amaya/
6 http://www.google.com/sidewiki/

4 Ruiz-Rube et al.
educational metadata approach that adapts existing video sequences as ready-to-
use learning resources. They exemplify the idea of referencing domain ontology
elements inside metadata elements, using Gene Ontology7 concepts to enrich
multimedia resources as part of the learning activity sequence. Time aspects of
such sequences are represented by using the MPEG-7 ontology [13], which classi-
es video as continuous annotated segments. Nevertheless, the MPEG-7 ontology
is uniquely linked to a set of predened concepts of multimedia resources (e.g.
dates and time), thus losing the genericity that may oer any W3C standard
ontology, such as TimeOntology8, for instance, to describe date-time relations
with the temporal dimension of elements of other domains.
3 Enriching Web Contents with Linked Data
We have built an approach to ease the task of enriching web content with lined
data by developing a plug-in that enables enriching text and video annotations
with embedded RDFa [1] that refer to concepts and relationships provided by
users. We have provided an extended interface to write/annotate texts, videos
and video fragments by customizing the open source TinyMCE9 with RDFa
capabilities. The reason of using RDFa is that web-based data and RDF an-
notations must be combined into the same web content container, because of
the great diversity of data applications to be linked (i.e. wikis, blogs, etc.) The
RDFa standard provides a high granularity for linked data modeling and can
be seamlessly integrated into web pages through XHTML attributes. It is at-
tainable and enables a cost-eective implementation [8]. Subsequently, Gleaning
Resource Descriptions from Dialects of Languages10 (GRDDL) can be used for
linking and extracting RDF data from the document.
During the development of our system, a number of semantic integration issues
have emerged. In the following, we rst dene such issues and afterwards propose
an approach to overcome them.
{ Generality vs Usability : One important issue to consider deals with the us-
ability of the annotation tool. Since our proposal is designed to be used with
any ontological model, its user interface must be completely decoupled from
that model. This feature, which a priori can be an advantage because of its
generality, might not be attractive enough to the end users, due to the lack
of the adaptation to particular domain knowledge. Therefore, a usability re-
view process would be required including the evaluation of real users (e.g.
by timing the tagging process spent by the users).
{ Annotation tracking : The blog entries, as any other content on the web, of-
ten evolve over time, i.e. once the contents have been created, they can be
updated several times or even be removed. It is common that annotations
7 http://www.geneontology.org/
8 http://www.w3.org/TR/owl-time/
9 http://tinymce.moxiecode.com/
10 http://www.w3.org/TR/grddl/

Development Issues on Linked Data Weblog Enrichment 5
existing at a given time cease to be valid, for example, by an error in the
annotation, which might cause a later update to the post. Thus, the ques-
tion about the persistence of annotations in the ontological model arises.
When should annotations be persisting? Possibilities include, among others,
the moment of annotation, on submitting the post, having an asynchronous
update task, on user demand, etc.
{ Annotation depth: RDFa recommendation allows to make annotations on
the
y within a given HTML content, i.e. to dene relationships between
concepts that are not stored yet and are visible on the post. This issue can
be solved by persisting annotations on the same moment of their provision.
However, it might provoke an inconsistency problem in the ontological model
if the post that is being annotated is not eventually saved. In consequence,
another solution for this issue is adding AJAX capabilities to the editor in
order to retrieve/save concepts that are not yet consolidated.
{ Annotation security and trustiness: Our proposal is intended to develop a
RDFa integrated editor that lets users annotate web contents using the exist-
ing concepts in the ontological model. This model is accessible to be queried
through a SPARQL11 endpoint. Typically, content managing systems |
as is our case with WordPress| provide authentication and authorization
mechanisms. However, the ontological data repository has to supply its own
security access mechanisms, since an authorized user can be able to mod-
ify blog contents and might not have enough privileges to work against the
repository.
Furthermore, one aspect to be considered occurs when two or more autho-
rized users are annotating the same concept and contradict themselves with
respect to the ontological model. This issue raises the need of an annotations
review process or a publication work
ow carried out by a specic domain
knowledge actor.
{ Reasoning vs performance: The content annotation process requires identi-
fying those concepts or concept types that can be used for annotations. It
also requires the establishment of relationships between them, according to
the axioms contained in the ontological model. The amount and complexity
of SPARQL queries necessary to support the user in this process will there-
fore depend on the complexity degree of the model itself. For example, in
complex models with dierent hierarchy levels, it requires the execution of
multiple SPARQL queries to collect all the data. This is originated by the
fact that the mechanisms to exploit the potential benets inherent to ontolo-
gies, such as class inheritance or inference, might be not natively provided
by the annotation system. This issue can be mitigated by using semantic
reasoners, which can infer logical consequences from a set of axioms and as-
serted facts from an ontology. Nonetheless, semantic reasoning comes at the
cost of signicantly decreasing performance.
{ Multiple source annotation models: Another issue arises when two or more
dierent ontology models are used as a reference of an annotation. For in-
11 http://www.w3.org/TR/rdf-sparql-protocol/

6 Ruiz-Rube et al.
stance, in order to carry out ontology-based annotations of the underlying
information held on a video sequence, the denition of intervals should be
explicitly linked with the multimedia resource. Furthermore, each of the
intervals dened within the multimedia resource must have a time represen-
tation (i.e. hasBeginning, hasEnd). This approach enables multiple semantic
embedded descriptions by linking each of the time intervals to concepts of
the domain ontology. To overcome this issue, an ontology that models these
dependencies is required, playing the role of 'semantic glue' between multi-
media resources, time intervals and user-dened concepts.
{ Complex ontology instances denition: When using certain ontologies, the
number of RDF statements involved to dene even simple concepts can com-
plicate RDFa annotations. For instance, OWL Time12 is a standard W3C
ontology for dening existing temporal concepts of Web pages and Web ser-
vices. This ontology model denes relations among instants and intervals,
together with information about durations and date-time information. In
order to annotate videos using time intervals, we can use OWL Time mod-
els. In the listing 1.1, we can see an example of how a time interval can be
described through the Time Ontology. However, the large number of RDF
statements and necessary RDFa code needed to dene a simple time interval
can make time annotations quite complex.
Listing 1.1. Sample Interval in OWL functional-style syntax
Main I n t e r v a l d e f i n i t i o n [1055]
ClassAssertion ( t ime : P rope r I n t e r va l : my In te rva l )
I n i t i a l i n s t a n t d e f i n i t i o n
ObjectPropertyAssertion ( t ime : hasBeginning : my In te rva l : i n s t a n t S t a r t )
ObjectPropertyAssertion ( t ime : inDateTime : i n s t a n t S t a r t : da teT imeDescr ip t ionSta r t )
ObjectPropertyAssertion ( t ime : uni tType : da teT imeDescr ip t ionSta r t t ime : unitSecond )
DataPropertyAssertion ( t ime : second : da teT imeDescr ip t ionSta r t ” 1 0 ” ˆ ˆ xsd : decimal )
F ina l i n s t a n t d e f i n i t i o n
ObjectPropertyAssertion ( t ime : hasEnd : my In te rva l : ins tantEnd )
ObjectPropertyAssertion ( t ime : inDateTime : ins tantEnd : dateTimeDescr ipt ionEnd )
ObjectPropertyAssertion ( t ime : uni tType : dateTimeDescr ipt ionEnd t ime : unitSecond )
DataPropertyAssertion ( t ime : second : dateTimeDescr ipt ionEnd ” 5 5 ” ˆ ˆ xsd : decimal )
To avoid the fact of having a complex amount of RDFa code generated by the
annotation tool, we have used a simplied way of dening OWL Time intervals.
With this solution, each multimedia resource is annotated with simple temporal
properties (e.g. start and end) and a RDF pre-harvesting process is done before
adding the RDF statements to the repository. This process is responsible for
translating the simple temporal properties into the proper OWL Time syntax.
4 Linked Blog application
The objective of LinkedBlog is to foster and enhance users' collaboration tasks
while examining and annotating the information held in a blog. Our proposal to
12 http://www.w3.org/TR/owl-time/

Development Issues on Linked Data Weblog Enrichment 7
metadata annotation would be helpful for users by granting them the possibility
of Linked Data editing. One objective of the Linked Data annotation approach is
to keep it independent from the web applications that are semantically enriched.
In order to eld-test this goal, we have extended the Wordpress blog engine with
an add-on used to enrich posts on the basis of concepts and relationships dened
by a standard ontology. Blog users can provide text and video annotations as
embedded RDFa. The second objective is to keep annotations independent from
the underlying ontology models used to describe web resources. For that rea-
son, the add-on can work with any ontological model available through a query
end-point compatible with the SPARQL protocol for RDF. There is only a de-
pendence of our approach with a specic ontology, but this is the Time Ontology,
that is general enough and serves to dene time intervals in video annotations.
4.1 Text and video annotations
LinkedBlog has been rst used13 in the context of the eCultura project [7],
which aims at providing a complete set of services and applications to access and
integrate diverse web-based contents of the cultural domain. These services and
applications use a data repository mainly fed with the CIDOC CRM ontology14
to depict cultural information, as well as with other ontologies, such as the Music
Ontology15, used to describe information about musical artists and works.
Text annotation The text annotating procedure is similar to, for instance, how
the user transforms a selected text into a hyperlink. Firstly, the user must select
a portion of text to process and then selects the concept. In the concept selector
pop-up window (see gure 1) the user can assign, change or delete a concept
(i.e. individual) and/or associate a concept type (i.e. class) to the enclosed text.
Once the user has selected the right concept or concept type, he/she is able
to insert or edit annotations. In the window that pops up (gure 2) the user
can indicate whether he/she wants to dene a direct relationship (i.e. an RDF
property having the concept as its domain) or a reverse relationship (i.e. an RDF
property having the concept as its range). Depending on the applicable range of
the relationship selected, the user can specify a literal content, the data type or
a related concept.
The listing 1.2 shows the HTML-embedded RDFa code that is generated
by the add-on after annotating a post about Chano Lobato
amenco singer. In
the example, you can see how his name is annotated with foaf:rstName (i.e.
Sebastian) and foaf:surname (i.e. Ramrez Sarabia) datatype properties, and his
discography using the mo:discography object property.
13 http://www.ecultura.org/blog
14 http://cidoc.ics.forth.gr/
15 http://musicontology.com/

8 Ruiz-Rube et al.
Fig. 1. Concept Selector
Fig. 2. Insert/Edit Annotation
Listing 1.2. RDFa generated code example for text annotation
<span about= ” h t t p : / / pu r l . org / onto logy /mo/ ChanoLobato ”
typeof= ” h t t p : / / pu r l . org / onto logy /mo/ So loMus icAr t i s t ”><span
re l= ” h t t p : / / xmlns . com/ f oa f / 0 . 1 / f i rs tName ”>Juan Miguel </ span> <span
re l= ” h t t p : / / xmlns . com/ f oa f / 0 . 1 / surname ”>Ramirez Sarabia</ span>(Chano Lobato ) was
born i n the Bar r i o de Santa Maria ( Cadiz ) , began in the flamenco stages of h i s
hometown , i n the Venta La Palma , w i th Au re l i o Sel les , Servando Roa and Antonio El
Herrero . Then he moved to Madrid to debut p ro f ess i ona l l y , becoming pa r t o f the
b a l l e t o f A le jandro Vega . Chano Lobato i n h i s <span
re l= ” h t t p : / / pu r l . org / onto logy /mo/ discography ”
resource= ” h t t p : / /www. deflamenco . com/ a r t i s t a s / ver . j sp?codigo=20 ”>discography</ span>
has more than ten albums .</ span>

Development Issues on Linked Data Weblog Enrichment 9
Video annotation LinkedBlog enables to annotate videos that are included in
a blog entry. The annotations can be done to videos of the Youtube platform,
because it is the most popular video provider that is integrated with Wordpress.
The procedure to describe the information held in the clip is as follows: First,
the user adds the video to the post. Once the video is added, the user must select
it (by mouse clicking) and then select the insert/edit video annotations option.
Fig. 3. Insert/Edit Video Annotations
For describing the information contained in the video, the user must dene
time intervals along it. For each interval, the user can specify the proper concept
and/or concept type, as well as the required annotations. To achieve this, the
user can set direct/indirect relationships with literals or other concepts.
Listing 1.3 shows the HTML-embedded RDFa code generated by the add-on
after annotating a blog post including a video about Chano Lobato and Juan
Carmona (also known as Juan Habichuela)
amenco artists, performing a Sola
(i.e. a kind of
amenco style). In this example, we have dened a single sample
interval, which describes the musical performance concept (mo:Performance).
There are also two mo:performed inverse relationships with Juan Carmona and
Chano Lobato artists (mo: MusicArtist). The RDFa snippet code includes both
concepts from the specied domain (in our case, MusicOntology) as well as the
needed concepts of the ontology used to specify time intervals.
Listing 1.3. RDFa generated code example for video annotation
<span about= ” h t t p : / /www. youtube . com/ v /RNbRNIwsY4 ”>
<span proper ty= ” h t t p : / /www.w3 . org /2006/ t ime# s t a r t ” content= ” 10 ”></ span>
<span proper ty= ” h t t p : / /www.w3 . org /2006/ t ime#end ” content= ” 50 ”></ span>
<span about= ” h t t p : / / pu r l . org / onto logy /mo/ Example ”
typeof= ” h t t p : / / pu r l . org / onto logy /mo/ Performance ”>
<span rev= ” h t t p : / / pu r l . org / onto logy /mo/ performed ”
resource= ” h t t p : / / pu r l . org / onto logy /mo/ ChanoLobato ”></ span>

10 Ruiz-Rube et al.
<span rev= ” h t t p : / / pu r l . org / onto logy /mo/ performed ”
resource= ” h t t p : / / pu r l . org / onto logy /mo/ JuanCarmona ”></ span>
</ span>
</ span>
4.2 Linked data support for semantic web
Web semantics provide reasoning capabilities that can help when querying on
complex domain models and inferring from a set of asserted class descriptions.
Using a semantic reasoner such as Pellet [17] enables to obtain all the properties
that can be applicable to a class (e.g. up to 86 properties formo:SoloMusicArtist
instances) by running a single SPARQL query, because the reasoner is able
to infer on the class hierarchy. Therefore, our add-on provides a specic end-
point interface for semantic reasoning enhanced queries. This implementation
drastically reduces the number and complexity of SPARQL queries.
Our add-on performs dierent SPARQL queries against the ontological repos-
itory. For example, when the insert/edit annotation window is shown, the repos-
itory will be asked by those object- or data-properties whose domain matches to
the current concept type or any of its supertypes.
In our case of study we use a repository based on MusicOntology (see Figure
4). When nding the applicable properties to mo:SoloMusicArtist, it is required
a pre-processing step that collects the set of nearby classes (i.e. foaf:Person and
mo:MusicArtist) and more distant superclasses (i.e. foaf:Agent). Subsequently,
the applicable properties to that set of classes are consulted.
Fig. 4. Excerpt from Music Ontology

Development Issues on Linked Data Weblog Enrichment 11
5 Conclusion and future work
In this paper, we described a set of issues that arise when developing an utility
to annotate text and videos in web contents. The main annotation issues that
must be addressed are the genericity vs usability, tracking, depth, security and
trustiness, reasoning vs performance, multiple source annotation models, and
complex ontology instances denition. There is an additional problem concerning
how it would be the best manner to identify concepts in the Web of Linked Data
and how these concepts should be organized in an URIs universal directory.
Nevertheless, the idea of managing a
exible and sustainable URI space is beyond
the scope of our work.
The LinkedBlog system enables to formally describe web contents using
Linked Data technologies, with independence of the underlying conceptual model
and loosely coupled with the application where it is hosted. The developed soft-
ware is an add-on that has been deployed in the Wordpress blog engine, but can
be easily integrated in other kind of content management systems.
In order to verify the strengths and the weaknesses of the add-on in a real
environment, we will do an implementation and a monitoring on the web portal
of the second edition of the university course for beginners about Flamenco
(www.
amencoenred.tv). In this domain, a specic enhancement is planned as a
future work to integrate Linked Data benets with music recognition capabilities,
either melodic [10] or rhythmic [11], in order to further automate the annotation
process of
amenco music themes.
Furthermore, it is argued that Humanities cannot perform their research
without ontology toolkits that handle ontological plurality and contexts, provide
argumentation and question corpora management systems, as well as support
adequate representational expressivity [15]. Therefore, the main evolution point
of our annotation approach is using the RDF reication mechanism. Thus, before
submitting any proposed semantic annotation to a server for persisting meta-
data, it can be contextualized for the corresponding user, post, and date-time.
Acknowledgments
This work has been sponsored by grants from the eCultura project (TSI-020501-
2008-53) of the Spanish Avanza R+D programme of the Ministry of Industry,
Tourism and Trade.
References
1. Adida, B., Hausenblas, M.: Rdfa use cases: Scenarios for embedding rdf in html.
World Wide Web Consortium, Working Draft WD-xhtml-rdfa-scenarios-20070330
(March 2007)
2. Allen, M.: Web 2.0: An argument against convergence. First Monday 13(3) (2008)
3. Berners-Lee, T., Fischetti, M.: Weaving the Web: The Original Design and Ulti-
mate Destiny of the World Wide Web by Its Inventor. Harper San Francisco (1999),
foreword By-Dertouzos, Michael L.

12 Ruiz-Rube et al.
4. Bizer, C., Heath, T., Berners-Lee, T.: Linked data{the story so far. International
Journal on Semantic Web and Information Systems 5(3), 1{22 (2009)
5. Bizer, C., Lehmann, J., Kobilarov, G., Auer, S., Becker, C., Cyganiak, R., Hell-
mann, S.: DBpedia-A crystallization point for the Web of Data. Web Semantics:
Science, Services and Agents on the World Wide Web (2009)
6. Bollacker, K., Evans, C., Paritosh, P., Sturge, T., Taylor, J.: Freebase: a collabora-
tively created graph database for structuring human knowledge. In: SIGMOD '08:
Proceedings of the 2008 ACM SIGMOD international conference on Management
of data. pp. 1247{1250. ACM, New York, NY, USA (2008)
7. Cornejo, C., Ruiz-Rube, I., Dodero, J.: eCultura, a Semantically-Enriched Web-
Based Approach to Manage Cultural Contents. In: International Conference on
Information Reuse and Integration (2010)
8. Dotsika, F.: Semantic APIs: Scaling up towards the semantic web. International
Journal of Information Management 30, 335{342 (2010)
9. Feng, S., Manmatha, R., Lavrenko, V.: Multiple bernoulli relevance models for
image and video annotation. In: CVPR (2). pp. 1002{1009 (2004)
10. Gomez, E., Bonada, J.: Automatic melodic transcription of
amenco singing. In:
4th Conf. on Interdisciplinary Musicology (CIM'08). Thessaloniki, Greece (2008)
11. Guastavino, C., Gomez, F., Toussaint, G., Marandola, F., Gomez, E.: Measuring
similarity between
amenco rhythmic patterns. Journal of New Music Research
38(2), 129{138 (2009)
12. Hirsch, C., Grundy, J., Hosking, J.: Thinkbase: A visual semantic wiki. In: 7th
International Semantic Web Conference (ISWC2008) (October 2008)
13. Hunter, J.: Adding multimedia to the semantic web: Building an mpeg-7 ontology.
In: SWWS. pp. 261{283 (2001)
14. Kahan, J., Koivunen, M.R., Prud'hommeaux, E., Swick, R.R.: Annotea: an open
RDF infrastructure for shared web annotations. Computer Networks 39(5), 589{
608 (2002)
15. Kahlert, R.C., Sullivan, J.: Ontologies: A wish list for the humanities. In: Proc. of
CaSTA'06 (2006)
16. Kiryakov, A., Popov, B., Terziev, I., Manov, D., Ognyano, D.: Semantic annota-
tion, indexing, and retrieval. J. Web Sem 2(1), 49{79 (2004)
17. Parsia, B., Sirin, E.: Pellet: An owl dl reasoner. In: Third International Semantic
Web Conference-Poster. Citeseer (2004)
18. Pinto, H., Martins, J.: A methodology for ontology integration. In: Proceedings of
the 1st international conference on Knowledge capture. p. 138. ACM (2001)
19. Schroeter, R., Hunter, J., Newman, A.: Annotating relationships between multiple
mixed-media digital objects by extending annotea. In: Proc. of the 4th European
Semantic Web Conference. pp. 533{548. Springer, Innsbruck, Austria (2007)
20. Sicilia, M.A., S.A.S., Lytras, M.: Semantic annotation of video fragments as learn-
ing objects: a case study with youtube videos and the gene ontology. Interactive
Learning Environments (2010, to appear)
21. Solc, T.: Automatic generation of in-text hyperlinks in web publishing. Tech. rep.,
Zemanta Ltd (2008)