Abstract— aceMedia has taken the challenge of organising
personal content by creating tools to automatically generate
metadata descriptors and search content intuitively. In this paper
we review part of the approach taken by aceMedia to create
semantic metadata (ontologies) and use of this to enable more
appropriate search and matching and managing of content.
However, there is further benefit the end user can gain from this
semantic metadata and this is from adding intelligence in the
software. The benefits and how intelligence can be added is
described with a particular focus on assisting the user in the
creation of privacy preference rules when sharing content: the
creation of self-governing inferencing.
Index Terms— semantic metadata, ontologies, intelligent
software, inferencing, self-governance. Topic “Integration of
multimedia processing and Semantic Web technologies (SS3)”
HE vision of the aceMedia project is to provide the tools
to assist in advanced content management. This is to deal
with the classic information overload, as users have not just
access to content in many forms but also many tools and
devices to create all types of content themselves.
Management of content becomes increasingly difficult such as
finding the right content, creating collections, annotating
aceMedia is researching methods to assist in information
and content management via knowledge technologies and
developments in the semantic web. The aceMedia approach
involves creating and using metadata to enable intelligent
applications such as advanced search and retrieval,
personalisation, self-organisation of content, and autonomous
content actions such as self-determined privacy. The use of
metadata does not come without some key challenges itself.
Many terms used within the metadata may refer to an implicit
informal semantics and do not necessarily provide essential
properties or relationships between terms to assist in any
automated approach to be applied. However, the move
towards the development of ontologies that model domains,
preferences, policies and profiles provide an approach to assist
in automating the matching and filtering of content searches.
Manuscript received June x, 2006. This work was supported in part by the
European Commission under contract FP6-001765 aceMedia.
P. Charlton and J. Teh are with Motorola Labs, Jays Close, Basingstoke,
RG22 4PD, United Kingdom. (e-mail: email@example.com).
II. ONTOLOGIES, METADATA AND SEMANTICS
Knowledge-assisted applications can enhance a user
experience because, for example, the user decides what they
want to access and can then set up an automatic feature that
the user controls. This is based on the same features that are
available to the application developer, because it is about the
same domain model; the only difference is in the access
perspective. The application developer can make available to
the user the same application configuration features that are
available to the application itself because of the formal
representation of the model. A granularity of control is made
possible because the formal model explicitly states what is
available. Further re-use is gained as the user profiles, models,
and data can be easily shared across many applications
III. ACEMEDIA PROJECT
aceMedia uses the metadata and ontologies to provide
context to assist in content management on behalf of the user.
The metadata and ontologies are used to:
Enable the user to search for content: the user can
enter natural language sentences that are parsed to
provide a structured query for the knowledge base.
Enable personalisation methods to create and manage
personal metadata to be applied to the search and
retrieval of appropriate content, in particular ranking
content, which is based on using machine learning
techniques to weight preferences and content.
Enable visual methods, which use intelligent
multimedia algorithms, through the low level
descriptors, to assist the user in matching similar
visual properties of particular content with other
As well as the above functionality to assist the user in
content management, aceMedia has researched into the
requirement and application of the intelligent layer, as part of
the Autonomous Content Entity (ACE). An ACE is a concept
which captures the content, metadata and the intelligence layer
as a type of intelligent media object. There are two specific
drivers that require the intelligence layer:
Digital Content is very nomadic; in that context it is
better to have content management attached to the
content such that the user can always optimally deal
with the content, wherever it resides
Digital content can easily flow to other places where
The Design of “Intelligence” for the
Management of Personal Multimedia Content
Patricia Charlton and Jonathan Teh
the owner of the content does not have control over Download full-text
this content; by carrying the rules of management with
the content then the owner’s rights and privacy
preferences are enabled.
The Intelligence Layer is defined as code executing in an
aceMedia system that provides “intelligence” to autonomously
support content management. This for example means how to
present itself, maintain and enhance the metadata, handle
privacy, self-adapt and self-organize etc. The intelligence
layer can be transported together with the content and
metadata as one object and the execution of the intelligence
layer is done in a secure environment.
There are two specific applications which make use of the
intelligence layer: personal content ownership rights to
support personal preferences about privacy of content through
self-governance and self-organising content to assist in the
automation of content collections dependent on the devices
and environments. The self-governance context is built on
assisting the user with a means to declare their access
preferences to their content. This in terms of concepts is close
to digital access rights, so access preferences available to the
user are captured in terms of digital rights access attributes.
A. aceMedia framework
The aceMedia system facilitates digital multimedia content
management through its software framework that enables the
execution of application modules and ACEs. The aceMedia
framework  enables ACEs to run, and reuse base content
analysis functionality, shared by all running ACEs. The
framework further enables users to control and restrain the
behaviour of ACEs. An application interface allows users to
interact with individual ACEs and manage ACE collections.
B. Intelligence layer: Requirements and Design
The intelligence layer provides a framework where code
(rules, methods and inferencing techniques) can exploit the
semantic structure of the explicit knowledge about the content
(multimedia content plus metadata descriptors, users, devices
The original concept of the ACE was about creating a
mechanism that would assist in the users in managing their
content and enable designers and developers methods to
support this managing of content while creating flexibility for:
a) tailoring solutions to the trends (ease of configuration) and
b) extensibility for future unknown development.
C. Design of the self-governance
In the user studies done by the aceMedia project there were
indications that the users had concerns and wanted to have
some control over their content and have certain checks about
how their personal content is shared and used . The
autonomous nature of content requires that the user’s intent is
carried with the content and this is captured and generated as
self-governing rules. The rules are generated dynamically
because it is impossible to define before hand all the possible
rules that user would consider when applying to their content
plus this gives the possibility to be extended to new contexts
that have not yet been considered. This approach to self-
governing inferencing system is further enabled because of the
structured semantic metadata used as grounding knowledge
within the overall aceMedia system. The concept of the
intelligence layer means that the rules can be executed when
the content is being accessed, used or modified in some way.
We capture the user’s privacy preference in a way that does
not demand constant attention from the user but does capture
the intended behaviour from the user by using a policy model
 to convert the user’s preferences into a rights context. This
rights context plus a priority system converts this into rules
that is carried with the context. There are default policies and
priority settings but these can be changed by the user and a
Some degree of self-governance of a computational system
is required if there is a need to support privacy preference-
based access to content when content leaves the control of the
owner. Here we express a degree of self-governance and the
self-governance of a system is required to be grounded in
some knowledge and facts about itself. Although theoretically
computational self-governance could be infinite, for example,
if the architecture incorporates a reflective tower principle 
e.g. of self-governing of the self-governing rules etc. but this
only indicates to the designer the flexibility potential of such
an architecture to deal with dynamics of a system itself.
We are now in the first stages of including into the
aceMedia framework computational intelligence. The use of
semantic metadata has provided some of the key cues for
enabling context-aware computations
computational intelligence functionality. It has provided the
semantic structure and grounding knowledge that is required
when performing intelligent behaviours.
The use of autonomy was drawn from the autonomic
computing of defining a notion of “self”. This is supported by
the aceMedia system: 1) the aceMedia framework enables a
concept of “code” within an ACE itself, 2) the semantic
metadata provide some context that can benefit and be
benefited by some form of computational intelligence, and 3)
the inclusion of the self-governing inference rules.
In providing the self-governing inference rules we have
added to the semantic metadata and have incorporated more
context cues through the use of policies that capture a user’s
privacy preferences. The semantic metadata has meant we can
define an inferencing system to build both the appropriate
privacy preference context and rules that can be attached to
the content creating an intelligence layer as part of the content.
 A. Matellanes, A. May, P. Villegas, F. Snijder, A. Kobzhev, E. O. Dijk,
“An architecture for multimedia content management”, EWIMT 2005.
 Smith B, “Reflection and Semantics in Lisp” ACM, pages 23-35, 1983.
 P. Charlton, J. Teh, A Self-governance Approach to Supporting Privacy
Preference-based Content Sharing, International Transactions on
Systems Science and Applications, Special issue 2006.