Often, one does not only want to segment a film into temporal and spatial segments using VIANs Segmentation, Screenshots or Annotations functionality, but also classify these by means of tags. The intention behind this is to structure the qualitative analysis using well defined sets of words, which allows querying and visualization of the collected data in a later stage. VIAN uses the term classification to describe this task, some would however call this tagging.
VIAN supports this using a techinique, called experiments. An Experiment simply encapsulates everything related to classification, making it possible to perform several unrelated classifications within one project.
In short, an Experiment can be described as follows: Once one has collected data in VIAN, that is, creating Segments, Stills or Annotations, one wants to classify this data by specific vocabularies. An experiment consists of a set of rules about how and which data should be classified. It defines which data should be classified by which vocabularies and which classification objects.
Following Example should elaborate the concept: Bob has just segmented a film called “Alice in Wonderland” into temporal segments in the segmentation called “Main Segmentation” and now wants to classify them by two attributes: If the color of a segment is saturated or desaturated and if the mediated emotion of the scene is happy, thrilling or sad.
In VIAN he would thus create two vocabularies: “Saturation” and “Emotion” “Saturation” contains two words: “saturated” and “desaturated”, “Emotion” contains three: “happy”, “thrilled” and “sad”.
Since Bob knows that the saturation of the protagonist’s dress is a completely different thing than the one of the supporting actors, he want’s to classify them seperately. He would thus create two classification objects “Protagonist” and “Support” and classify both by the vocabulary “Saturation”. He then creates a third Classification Object “Global” for the emotion since it is not focused on the protagonist in his case.
Finally he would set for all three classification objects to target his “Main Segmentation”. VIAN will now create a questionnaire for all segments, which he can classify.
To create a new Experiment, do the following:
Go to Create/Create Experiment.
In the outliner select the new experiment.
In the experiment editor open the General tab to rename the experiment to your needs.
Setting up an experiment can feel a bit complicated at the beginning, but will become more intuitive once you understood, VIAN’s way of looking at things.
Once you have configured your experiment as described in _creating_experiments you can edit the experiment in the experiment editor. You should now see this widget:
To add classification objects, do the following:
In the Classification tab, add a classification object by typing its name into the textbox on the bottom.
On the right side, select the Vocabularies tab.
In Attached Vocabularies tick all vocabularies with which you want to classify the created classification object.
Classification Objects A classification object may be both a conceptual or true object which you want to classify, and may have a pixel representation within the film material. E.g. A classification object could be named Figure or Ground, since only the Figure can have an emotional state, the vocabulary Emotion would only be used to classify Figure. A conceptual classification object could for example be Soundtrack.