Film Production Ontology

A film production ontology has been developed in ANSWER to provide a semantic framework for Director Notation and for other metadata created during production planning. The ontology has two distinct components, each of which forms an ontology in its own right. The Core Ontology is application specific and covers the concepts necessary to populate the scene description using the DirectorNotation. The Context Ontology covers the broader aspects of film and digital media and will be extended to encompass the concepts from the games domain. The ontology is summarised below, but a richly-illustrated detailed description can be found in the corresponding project deliverable.

The Core Ontology covers the DirectorNotation specific concepts. Its primary technical purpose is the translation of input notation into output representations such as PML (Player Markup Language). It is also a formalisation of the artistic language itself - thus it is intended that there can be no dispute at an artistic level about what a given script written in notation is intended to mean, as the language's semantics are formally defined. DirectorNotation is a specification of a conceptualisation. It defines specific, concrete concepts as well as their interrelation, such that a (syntactically correct) instantiation of these concepts and relations describes the content of a cinematic work. Therefore, a simple one-to-one representation in a technical, formal language of DN's symbols as concepts and of their syntactic connections as relations is an ontology by definition. However, such an ontology is not an actionable semantic model.

The queries that we wish to evaluate involve finding unknown quantities that do not derive from the syntactic rules of the notation, but follow from a process of reasoning about the directorial concepts represented by the symbols, and the interactions of these concepts. Thus, the symbols of the notation represent directing concepts, but superficial reasoning at the level of symbols only is meaningless, as the human expert reasons about the underlying concepts and merely represents the initial conditions and the conclusions arrived at in the language of notation. A significant part of the Director's knowledge is understanding and capability of the expert that is hidden in their skills and that they rely on fundamentally when reasoning about the ideas she has expressed in notation. We draw on this knowledge to explicitly model this expert knowledge. This support structure will enable automated reasoning with the specific capability of answering queries about notationdescribed-content. In other words, this structure completes our knowledge representation so that the semantics can be considered actionable, once additional evaluation algorithms have also been implemented.

In order that Director Notation be actionable, the Core Ontology thus models:

  • firstly, the transition from symbolic representation to underlying directing concepts;
  • secondly, the possible conclusions about directing concepts (i.e. directing concepts are linked to each other representing the workflow of reasoning about them);
  • and thirdly, the physical, geometric knowledge that is also required in order to understand film.


Correspondingly, there are three main levels in the Core Ontology:
  • Notation layer: mapping to input data (e.g. to the notation editor's XML schema);
  • Directing concepts layer: mapping between the Notation layer and the Geometry layer;
  • Film Geometry layer.



The Context Ontology covers a broad range of concepts, from abstract ideas like film genre and mood, to production specifics including locations and camera equipment. Key top-level classes include Theatrical Presentation Element, Physical Entity and Design Element. Theatrical Presentation Elements include Shot Angles, Camera Angles, Camera and Lighting Types etc.

Examples of Physical Entity that are used during a film production process include Lighting Equipment, Optical Attachments, the Film Script, the Camera etc. Design Elements are a combination of various narrative and dramatic elements. The design element may either be used during Screenplay or during Production. Examples include Sound, Costume, Props etc.

SWRL (Semantic Web Reasoning Language) rules are included in the ontology, allowing simple inference to take place. For example, a camera lens can be classified automatically according to its physical specification. Also modelled in the context ontology are aspects of film grammar - standard conventions followed by directors while shooting a scene. Our intention is that, ultimately, by reasoning over film grammar codified in the ontology, the ANSWER rule engine will be able to ensure both the syntactic and the semantic validity of notation scores. An example of a rule used in film production is the 30 Degree rule: A shot of an object from a particular camera angle X in a time frame T should change by at least 30 degrees angle if the same object is being shown in the next frame.

In developing the Context Ontology we have been mindful of other work in this space, and have sought to reuse existing ontologies where appropriate. Notably, we have re-used the MPEG-7 upper level ontology as the foundational ontology. MPEG7 is widely used; it provides support to a broad range of applications, ranging from Multimedia digital libraries, broadcast media selection, multimedia editing, home entertainment devices etc. Further it also covers many of the specific aspects related to content description such as Colour and Texture Descriptors, support for 2D and 3D structures and Motion Descriptors which describe the Camera Motion, Motion Trajectories and Motion Activity.

Development of the ANSWER Context Ontology has taken place in two stages. Initial design work was informed by a study of film production texts and collaborative sessions with external experts. Further collaborative refinement and extension of the ontology is now underway between the project partners; for this we are making use of the Collaborative Protégé platform.