Semantic Authoring By Tagging with
Annotea Social Bookmarks and Topics
Marja-Riitta Koivunen
Annotea
marja@annotea.org
ABSTRACT
Annotea is a framework that supports Semantic Web based
1. INTRODUCTION
Annotea is a framework for semantic annotations [26],
collaboration and authoring via Annotea objects, such as social
annotations, bookmarks, and topics. Social bookmarks and bookmarks, and topics focusing on sharing of the objects and
collaboration in document context[18]. Annotea [4, 17] gets its
topics can be used for semantic authoring in many levels. They
inspiration from observing users’ needs and problems and
let ordinary users tag interesting Web documents with their own
personal concepts (folksonomies). These concepts can be simple developing user scenarios based on those observations [18].
Annotea sees humans as meaning makers. They create concepts,
or form hierarchies, and they can be linked to more standard
concepts when users become aware of them. Collections of the make associations, comment, clarify, annotate, review, organize,
bookmark, tag, and file information into categories or organize it
concepts and the bookmarks under them can be stored to
bookmark documents or services as metadata and easily shared into folders. When humans learn something new they augment
and modify their existing cognitive models with new personal
with other users. Annotea users can also literally semantically
author pages with link collections and categories. They can use concepts. Typically, these concepts are vague at first but as they
become more precise they can be shared with other users and
the title and description information of the topics and
developed further collaboratively.
bookmarks to create the page content. This content from the
bookmark and topics hierarchies is then easily converted to Ordinary users who are not familiar with semantic Web can do
semantic authoring with intuitively familiar Annotea objects,
XHTML Web pages by using XSLT scripts. Annotea
presentation objects further support authoring and customized such as annotations, discussions threads, bookmarks, and topics
(tags). For instance, Annotea social bookmarks allow users to
presentations by separating the presentation and the application
data. While Annotea precedes tagging and social bookmarking recall earlier visited Web resources, such as Web documents,
services, music, images and blogs. As a side effect, the user also
services and tools popular today it still provides better
authors metadata about the bookmarked resources. When
flexibility, extendibility, and reuse of the bookmark and topic
data with help of the Semantic Web technologies. published and shared, this information can be useful not only for
the user but also for other users and applications.
Users can also choose to use Annotea to author and maintain
Categories and Subject Descriptors Web pages consisting of link collections and categories. In this
H.5.3 [Group and Organization Interfaces]: Computer-
case, users semantically author bookmark and topic information,
supported Cooperative Work, Web-based Interaction; I.2.4 such as title and description information, which can be used to
[Knowledge Representation Formalisms and Methods]:
form Web pages that were earlier manually written and
Semantic Networks –Semantic Web maintained.
Annotea topics [18, 19, 21, 16] support user defined simple
General Terms concepts (tags). Instead of forcing the users to invest in learning
Design, Human Factors standard ontologies before they can start developing their ideas,
Annotea supports the innovation process by letting users start
Keywords from their own personal ontologies (folksonomies). Gradually,
when the user learns about connections her concepts and
Social bookmarks, social topics, Semantic Web, standard, commonly used concepts, it is possible to link the
semantic authoring, ontologies, tagging, folksonomies. personal topics to them.
Annotea uses the Semantic Web technologies for implementing
the solutions but hides the Semantic Web from the ordinary
users under the familiar objects and easy user interfaces. From
the user point of view, the Annotea topics can be used for Web
tagging just like in the popular Web tagging services [11, 13]
that followed Annotea. However, the Semantic Web metadata
basis [9] makes Annotea more flexible, easily extendible, and
interoperable supporting user needs and applications also
beyond tagging. The user can select bookmark stores relevant to
her tasks and seamlessly merge Annotea objects from these document can also be used to archive snapshots of the selected
stores to support task related queries. Annotea objects stored on the server. For instance, a version of a
Annotea objects can be presented to the users in many document can be stored together with annotations for that
innovative ways and by different tools. Typically, the objects are version.
explored in the context of the Web resources they refer to, Annotea objects have URIs and properties. They all store some
examined in selected topic hierarchies, and searched by using basic tracking information, such as who created them and when.
different properties. Amaya browser/editor [1] and lately The Annotea schemas [2, 3, 6] define only a few specifically
Annotea Ubimarks [7] have been used to demonstrate some Annotea related properties for Annotea objects. To improve
capabilities of Annotea objects. In addition, many other clients, reuse and decrease unnecessary inferencing the rest of the
such as Annozilla [8], have been developed outside the Annotea properties are adopted from other well-established schemas,
team. The bookmark files can also be presented in normal such as Dublin Core. Annotea usage documents [20] and server
browsers as XHTML pages by using selected XSLT style sheets. protocols [25] help define the basic annotation/bookmarking
In Annotea the bookmarks and topics can be stored locally or framework [14] for Annotea objects. Finally, model user
shared both with individual users on different devices and with interfaces [1,7] demonstrate the use of the objects.
different user groups. Unlike simple tagging or bookmark New properties can be easily added to Annotea objects. For
formats, such as XBEL [27], Annotea bookmarks and topics instance, Annotea Ubimarks lets the author add a URI to his
help collaboration and finding relevant information even when personal and social network information by using foaf:maker
users use different terms for similar concepts. This is is often the (see Figure 1). When provided, this social network information
case in cross disciplinary fields, such as life science, user can be used as additional help when trying to find related
experience or even in marketing and research where users are documents.
awarded from finding original concepts instead of using the Annotea [14 , 17] has demonstrated storing annotations and
already existing ones. While services, such as Linkagogo [22] replies to different annotation servers. Annotea protocol for the
help in interpreting between different formats, the simple metadata stores is based on HTTP. Annotea bookmarks and
formats often don’t have enough identification information to topics can use a similar protocol or standard Web queries. We
easily separate or merge concepts.
have done some experiments with this. Currently, the model
In addition to supporting tools for creating and retrieving implementations use HTTP, FTP, and possibly also other Web
bookmarks, the Annotea bookmarks can be used as user profiles protocols to store the bookmarks and topics as Web documents.
for different services. This helps users find the information This approach can be applied for annotations and replies as well.
relevant to them at a certain context. Annotea bookmarks and
Annotea applications and user interfaces provide different
topics can also contain information not just about the user but views of the Annotea objects. When the interfaces are attached
his or her social network. Again, this can be used to support the to browsers users can seamlessly collaborate via Annotea
finding of more relevant information. Bookmarks and topics can objects in the context of the Web documents or other Web
also be used in applications organizing blog type information. resources. In addition, Annotea objects can be used in many
Finally, as shared bookmarks are easy to create we hope to other kinds of applications and user interfaces. For instance,
eventually see enough commonly available bookmark and topic bookmark collections on a selected topic can be used as user
data to be able to see network effects in applications using the defined profiles for services or bookmark repositories for
published bookmark repositories collectively. These selected domains can be used to order the results from search
applications can provide more relevant results. Furthermore, agents, such as Yahoo Search or Google, in a more meaningful
when dat a mining and search engines type applications mine the way for the user [24].
topics and different connections between the topics it is possible
to make many new innovative discoveries connecting fields that 3. ANNOTEA BOOKMARK AND TOPIC
were not aware of each other because they used different OBJECTS
terminologies. While the first phase of Annotea development concentrated on
annotations, early on many user scenarios were developed,
2. ANNOTEA ARCHITECTURE presented, and discussed about social bookmarks and topics
Different annotation and tagging applications can be built on top (tags). These were implemented during the second development
of Annotea architecture that includes Annotea stores, Annotea phase, when the need for categories was high on the users’ wish
objects, such as the bookmarks and topics, and protocols. list. Especially status categories were needed to mark the
Annotea stores are RDF metadata stores that store Annotea processed annotations.
objects. Objects from various stores can be easily merged and We wanted to make sure that Annotea could easily support not
viewed together. Many currently popular tagging services only annotation categories but also our other scenarios related to
violate the principle of user control and do not give users a categories. We also wanted an easy user metaphor that would
choice of where to store the bookmarks and topics. The Annotea work well with the already existing metaphors. As a result, we
objects can be stored in local files, global Web servers and/or decided to implement the social bookmark and topic objects as a
Web documents containing the metadata. variation of the semantic annotations [26]. Our goal was to
Web documents offer users an easy alternative to get started experiment with these objects and use the experiences later to
without a server. The same authentication and protection also improve the annotation implementations.
mechanisms can be used as for any other Web document. A
Figure 1. A sample Annotea bookmark and topic as RDF.
The bookmark metaphor was great for our purposes. Ordinary Alternatively, the user can choose HTTP addresses in her own
users are not only familiar with it but have often used traditional domain with date info by defining an URI path, for instance,
bookmark interfaces. Furthermore, the bookmark user interfaces "http://mydomain.org/AnnoteaURI/". Similarly, a
have a lot of enhancement possibilities that benefit from the bookmark server can create URIs. If the URI changes in the
Semantic Web approach. For instance, the user interface can server or for some other reason, the correspondence between
utilize the document context to find similar bookmarks and two Annotea objects can be defined by using the owl:sameAs
topics.. property.
3.1 Annotea Object Identification 3.2 Bookmark and Topic Properties
Like other Annotea objects, the bookmark and the topic objects Annotea does not define all bookmark and topic properties by
have their own identification URI. This URI ensures that itself. It uses properties from well known vocabularies, such as
Annotea objects retain their identity globally even when objects Dublin Core. Applications already familiar with these properties
are merged together from different data stores, or moved to know how to deal with them also in Annotea object context
another location along with a bookmark file, or sent to another without any extra processing. RDF for a sample bookmark and a
user by e-mail. topic is presented in Figure 1.
Not all users have their own Web domains. Therefore, the Bookmarks and topics have many common properties with other
current Annotea Ubimarks implementation uses URIs that are Annotea objects. For instance, the objects can have a title
unique UUIDs, such as "urn:uid:025f3d84-57bd-11da-924d- (dc:title), a longer description (dc:description), information
b619a79a047d". The UUIDs can be easily created without about their author (dc:creator and foaf:maker), the creation date
owning a Web domain. Resolving the objects content based on (a:created), the modification date (dc:date), and links to other
the UUID would need more work, but that is not usually a related information (RDFS:seeAlso)
problem. .
Figure 2. Clicking properties icon to open a bookmark view with common bookmark properties in Annotea Ubimarks.
A topic view is presented in the background.
In addition, a bookmark has information about the page it recalls concepts in a better established ontologies, her own concepts
(bm:recalls), and zero or more topics (bm:hasTopic). A can be linked to those.
topic has information about its parent topic
(bm:subTopicOf). In addition, new properties can be easily 4.1 Bookmark and Topic Object View
added. Users can explore the bookmark or the topic information in
simple forms. For instance, in Figure 2 the “Bookmark view”
4. VIEWING ANNOTEA BOOKMARKS presents the bookmark’s RDF properties and values.
AND TOPICS Properties can be added to Annotea objects and different
Annotea objects can be presented to users without a need to presentations may show different sets of properties. With
know anything about the Semantic Web. The bookmark guidance information about the properties, such as the
metaphor nicely supports semantic authoring and presentation of presentation order and the preferred widgets, the presentations
Annotea bookmark and topic objects, their collections, and could be created automatically. This is the next step for the new
context sensitive views to these objects. Annotea presentation objects.
The Annotea topics can be anything the user wants them to be.
They can start with vague meaning and gradually become more
robust and better understood. When user discovers other related
Figure 3. Annotea related projects page started as a manually edited Web page (bottom window). Next it evolved into a Semantic
Web based Annotea bookmark hierarchy maintained with Annotea Ubimarks (middle window ). Finally, the bookmark hierarchy
semantics was presented as a Web page by using XSLT transformation (top window) making the original page obsolete.
“Annotea related projects”. Clicking the bookmark title in the
4.2 Viewing Collections of Bookmarks and hierarchy view shows the recalled page while clicking the
Topics properties icon presents the bookmark view (see Figure 2).
Annotea topics allow authors create and maintain shared Clicking a topic always shows its properties in a topic view.
classifications or topics [15]. A bookmark can be cataloged
under one or more topics and presented to the user in a The Semantic information in Figure 3 bookmark hierarchy was
originally copied from the Web page shown in the bottom
hierarchy view resembling a folder hierarchy. In Figure 3 the
browser window. The bookmark document was so easy to
“View Bookmark Hierarchy” window shows a hierarchy of
update with Annotea Ubimarks that in many cases the updates to
the original Web page lacked behind. Fortunately, it was quite bookmarked the current page. The “Pagemarks” icon can be
easy to create a corresponding XHTML page directly from the clicked to open a view listing all the bookmarks on the page.
bookmark document semantics with an XSLT transformation. These bookmarks and their topics can be easily followed to find
The top browser window in Figure 3 shows one example of related documents in the currently subscribed bookmark stores,
such a transformed page. The style of the generated page can be some of which may provide information from Web crawlers.
easily changed to bear even closer resemblance to the original Users collaborating in related research areas can benefit from
page. This is semantic authoring at its best as the user gets the bookmark and topic information. They can find related
benefits from semantics that was created without any extra work topics and bookmarks. For instance, in Figure 4 the user has
or learning. bookmarked a gene page in Amigo. When she browses the page
later and clicks the “Pagemarks” and the topic icons, she finds
4.3 Viewing Bookmarks and Topics in more information about the genes involved in CML leukemia
Context made by other researchers in other departments of the institution
Context dependent views of Annotea objects are characteristic or in other institutions. In the future, the user generated
to Annotea applications. For instance, Annotea Ubimarks user information can be combined with automatic categorizations,
interface provides information about related Web documents in such as created in MagPie [12] or in Trailblazer [10] to help the
the current document context. When a user browses through the user to find also non-bookmarked relevant documents and make
Web pages and finds something interesting she can look for the new innovative associations.
“Pagemarks” icon. This icon means that one or more users has
Figure 4. Following the chain from a pagemark and its topic to related bookmarks.
Figure 5. A simple Annotea presentation object.
Workflow status (see Figure 6) and content label hierarchies
4.4 Defining Custom Bookmark Orders (Figure 7) are such examples.
The order of the bookmarks can be defined by using Annotea
presentation objects [5]. They can be attached to another 5. CONCLUSIONS
resource, such as a topic, to guide its presentation. For instance,
Annotea has been pioneering with social bookmarking, tagging,
it can define the order of the selected bookmarks under a topic. and folksonomy applications on the Web. Annotea social
The rest of the topics are presented alphabetically. The
bookmarks and topics provide ordinary users familiar metaphors
presentation objects let users define customized orders for a for authoring and sharing reusable and easily extendible
shared hierarchy. A sample presentation object is presented in
Semantic Web metadata without a need to learn or understand
Figure 5. the Semantic Web.
Many Web pages containing links and categories can be chosen
to be semantically authored as Annotea bookmark files and
transformed with XSLT into XHTML pages. This makes
maintaining the links easy.
Annotea users are not limited to use one service for their
bookmarks and topics. They can subscribe several local or
global data stores containing the various Annotea objects they
are interested in at each moment. Web documents offer users an
easy starting point. Semantic Web gives supports for easy
merging and querying of Annotea metadata from several
different sources.
Annotea object metadata is uniquely identifiable and includes
information of its creator and creation dates. It can be easily
extended, customized, combined, and reused in many
applications and services. It can provide additional organization
Figure 6. Using topics for attaching status values for for data mining and search engine applications, chatrooms, and
bookmarked problem descriptions. blogs.
Different views to the data, including document context
dependent views, can be easily created. Overall, Annotea
bookmarks and topics can provide huge benefits and flexibility
for Semantic authoring without adding complexity for the user
interface.
6. ACKNOWLEDGMENTS
Many people including Jose Kahan, Eric Prud'hommeaux
and Ralph Swick have contributed to Annotea.. I also
want to thank Siegfried Handshuh, Martin Dzbor and Pat
Croke for support and many inspiring discussions.
7. REFERENCES
[1] Amaya homepage, http://www.w3.org/Amaya
[2] Annotea annotation schema,
http://www.w3.org/2000/10/annotation-ns
[3] Annotea bookmark schema,
http://www.w3.org/2002/01/bookmark#
Figure 7. Using topics for content labels. [4] Annotea homepage at W3C,
http://www.w3.org/2001/Annotea
[5] Annotea presentation schema (draft),
4.5 Defining Multiple Root Topics http://www.annotea.org/schemas/2006/04/presentation
The topic hierarchies do not have to have a common root topic.
It is possible to have several roots. This lets the user also view [6] Annotea thread schema, http://www.w3.org/2001/03/thread
hierarchies even when some topics are temporarily unavailable. [7] Annotea Ubimarks homepage,
Hierarchies containing several root topics can be also viewed in http://www.annotea.org/mozilla/ubi.html
multifaceted displays, such as the ones used in mSpace [23].
[8] Annozilla home page, http://annozilla.mozdev.org/
Because of multiple roots it is easy to define concept hierarchies
that are intended to be used in parallel with other concepts.
[9] Brickley, D., and Guha R.V. (eds.). RDF Vocabulary KCAP 2001 Conference,
Description Language 1.0: RDF Schema, W3C http://www.w3.org/2001/Annotea/Papers/KCAP01/annotea
Recommendation 10 February 2004. .html
http://www.w3.org/TR/2004/REC-rdfschema-20040210/ [19] Koivunen, M., Swick, R., Kahan, J., Prud'Hommeaux, E.,
[10] Croke, P., Johnston, A., Tighe, K., Using Named Entities Topics in Annotea: Extending the Basic
as a basis to share associative trails between Semantic Architecture,.WWW2002 Developers Day, Honolulu, May
Desktops, 2002.
http://www.semanticdesktop.org/xwiki/bin/download/Wiki/ [20] Koivunen, M., Swick, R., Kahan, J., Prud'hommeaux, E.,
UsingNamedEntitiesAsABasisToShareAssociativeTrailsBe An Annotea Bookmark Schema, 2003,
tweenSemanticDesktops/19_croke_jack_final.pdf http://www.w3.org/2003/07/Annotea/BookmarkSchema-
[11] Del.icio.us homepage, http://del.icio.us/ 20030707
[12] Domingue, J.B., Dzbor, M., Motta, E. (2004) Collaborative [21] Koivunen, M., Swick, R., and Prud'hommeaux, E. Annotea
Semantic Web Browsing with Magpie, In Proc. of the 1st Shared Bookmarks, 2003, In Proc. of KCAP 2003,
European Semantic Web Symposium (ESWS), May 2004, http://www.w3.org/2001/Annotea/Papers/KCAP03/annotea
Greece. bm.html
http://kmi.open.ac.uk/people/dzbor/public/2004/iswc2004- [22] Linkagogo homepage, http://www.linkagogo.com/
dzbor-etal.pdf
[23] Schraefel, M.C., Carr, L., De Roure, D., et al., You've Got
[13] Flickr homepage, http://www.flickr.com Hypertext. Journal of Digital Information, 2004. 5(1): id
[14] Kahan, J., Koivunen, M., Prud'Hommeaux, E., and Swick, 253.
R. Annotea: An Open RDF Infrastructure for Shared Web [24] Shiraishi, N. (2004) The RDF Trust Model Using RDF
Annotations, in Proc. of the WWW10 International Bookmark and it's Application. In Proc. of WWW2004
Conference, Hong Kong, May 2001 Workshop on Content Labeling -Technical and Socio-
http://www10.org/cdrom/papers/488/index.html. Cultural Challenges and Solutions.
[15] Koivunen, M. Scenario: Organizing CML cancer research http://web.sfc.keio.ac.jp/~kaz/www2004/papers/ns.pdf
knowledge by using Annotea shared bookmarks, 2003, [25] Swick, R., Prud’Hommeaux, E., Koivunen, M., Kahan, J.,
http://www.w3.org/2003/12/cmlcase/cml.html Annotea protocols,
[16] Koivunen, M., Annotea shared bookmarks: Semantic Web http://www.w3.org/2001/Annotea/User/Protocol.html.
at your fingertips, In Proc. Of the ISWC 2004 Conference [26] Uren, V.S., Cimiano, P., Iria, J., et al., Semantic
Demonstrations Session. Annotation for Knowledge Management: Requirements
http://www.annotea.org/ISWC2004/annoteademo.html and a Survey of the State of the Art. Journal of Web
[17] Koivunen, M. Annotea and Semantic Web Supported Semantics, 2006. 4(1): p.14-28.
Collaboration, in Proc. Of the ESWC2005 Conference, http://www.websemanticsjournal.org/ps/pub/showDoc.Fullt
Crete, May 2005. ext/document.pdf?lang=en&doc=2005-
http://kmi.open.ac.uk/events/usersweb/papers/01_koivunen 34&format=pdf&compression=
_final.pdf [27] XML Bookmark Exchange Language,
[18] Koivunen, M. and Swick, R. Metadata Based Annotation http://pyxml.sourceforge.net/topics/xbel/
Infrastructure offers Flexibility and Extensibility for
Collaborative Applications and Beyond, In Proc. of the