- In this short paper we report on a semantic model for content and knowledge which distinguishes between three descriptive levels: information relating directly to the resource, to the meta data of the resource, and to the subject matter addressed by the content. This model addresses five fundamental requirements for automation: formality, interoperability, multiple interpretations, contextualisation, and independence of knowledge items from the resource's content.
Index Terms— knowledge content objects, intelligent content models, rich media content.
Semantics – i.e. the interpretation of the content – is
important to make content machine-processable and to enable
the definition of tasks in workflow-environments for
knowledge workers in the content industries. Some of the recent
research projects in the area of semantic (or symbolic) video
annotation try to derive the semantics from the videos’ low
level features or from other available basic metadata. Most
of these approaches are – as also pointed out in [
We suggest that representing richer semantics for multimedia content requires more expressive and more sophisticated knowledge content models than those currently used. We therefore introduce a model for representing knowledge and content alongside each other, with clear separation of the content and the knowledge items, so as to obtain optimal conditions for content and knowledge reuse, and for subsequent re-contextualization of content.
For a long time, combining knowledge and content did not
play a great role in the research communities: On the one hand
there were metadata models for content like MPEG-7 [
Possible relations between knowledge and content are
manifold. Based on observed applications and requirements of
current projects (see [
KCOs are based on the DOLCE foundational ontology5 and have so-called semantic facets that form modular entities to describe the properties of KCOs, including the raw content object or media file, metadata and knowledge specific to the content object and knowledge about the topics of the content (its meaning).
Knowledge is represented by the structure of the KCO in three different levels: 1) Resource Level: This level refers to the actual content object (File, stream, image, etc). 2) Meta Level: This level refers to knowledge describing features of the content object, eg. frame rate, compression type or colour coding scheme. 3) Subject Matter Level: This level comprises knowledge about the topic (subject) of the content as interpreted by an actor. The content object realizes this interpretation. 3http://www.acemedia.org 4http://metokis.salzburgresearch.at 5http://www.loa-cnr.it/DOLCE.html
In addition to this knowledge structure the KCO also defines
a structure based on the different domains of the knowledge
objects. This structure is divided into six so-called facets, each
of them optimized for a specific usage. Facets include for
example a content- or community description facet [
Relating the above description to the requirements from section III on knowledge content, we suggest that KCOs provide a good foundation for modeling combinations of content and knowledge.
1) Knowledge must be encoded using a formal language: KCOs are based on the information objects design pattern, which is an extension of the DOLCE foundational ontology. The main concepts and relations used in the description of the KCO are well grounded on this foundational framework. The current definition of KCOs is based on OWL-DL6. 2) Interoperability especially for cross domain aspects: The facet based structure of the KCO serves as a good starting point for the alignment of standards to the KCO structure. Some of the facets and elements there use parts of different standards like NewsML7 or MPEG-7. 3) Different interpretations of content objects: The possibility of different interpretations of content objects is the main reason for distinguishing between the mesta level and the subject matter level. Thus the KCO model support multiple interpretations of one content object. 4) Clear definition of possible relations between content and knowledge: The KCO defines two different interrelations between content and knowledge: First, knowledge objects can be about a content object, meaning that the subject matter of the knowledge object is the content object itself. Second, knowledge objects can be realized by content objects. This relation is used for all knowledge objects which are about the subject matter of the content object. 5) Make knowledge independent of content: This requirement is modeled by the fact that knowledge objects which belong to the subject matter level are about an arbitrary topic and only realized by the actual content.
In this section we will shortly sketch how the ideas of KCOs can be applied to address the various conceptual relationships between content and knowledge in media-rich systems.
a) Search based on meta data: KCOs can be used to model complex queries: In mental models that are representing users intentions, subject matter level information about the topic of the content is often mixed up with meta level information about the content objects: By posing queries to the system, users create descriptions about topics or subjects that they are interested in.
b) Collaborative filtering: A query in that setting specifies the actual context of the actor by considering some of the concepts and relations within the knowledge space of the actor, which are typically stored in the form of user profiles containing additional knowledge about preferences of the users. This information can be used to further contextualize queries by combining the context specified by the query, with the characteristics of the user profile. Based on the active concepts and relations of the contextualized query, the system can find similar interpretations. Such a system is sensitive to different interpretations of one and the same content object, because it handles different interpretations of different users separately.
c) Context-based content classification to minimise the semantic gap: This scenario refers to the problem of how to overcome the gap between low level features and higher level semantics. It assumes that new content objects typically have to be analysed at the expense of some time and effort. The complextity of this operation can be reduced based on background information about the content or some predefined knowledge of parts of the content as knowledge about one part can help to understand other parts of the content.
More details about the reported work can be found in [