Work with Advene: how this package was constructed

This document is a summary of the construction of the package nelson.xml: schema design, annotations, view design.

First step: a short summary of the video

We have defined a first schema, named Summary, that contains two annotation types: PartOfSpeech (text/plain) and TalkDescription (application/x-advene-structured) with speaker, date, place, event attributes. We also defined a relation type : NextPart (text/plain)

We have the defined annotations: first one annotation of type TalkDescription, then 11 annotations of type PartOfSpeech. These latter cover all the video, and annotation was done using the "on the fly" annotation mode from the main Advene view, then adjusted with the timeline view time setting facilities. 11 relations have also been defined between the annotations, the last one being linked to the first.

We then defined

• a static view named S_Summary which presents summary informations (from the 1 + 11 annotations)
• a dynamic view named D_Summary that contains 3 rules:
  • caption of the title of the PartOfSpeech
  • push the URL of the part on the URL stack
  • navigation popup with two links, one to the next part of speech, the other to the previous part of speech, taking advantage of the nextPart relatio

Second step: linking transcription to the video

We first made a transcription of Ted Nelson speech.

Then we linked the transcription to the video, which means:

• Defining a new schema DiscourseAnalysis.
• Defining a textual annotation type utterance.
• Synchronizing with the video using the import transcription GUI. The chosen granularity has the "subtitling" use in mind : short utterances, but not too short. The interested reader can download the transcription import work saved file, and open it into the import transcription GUI (here is a shortcut).

Some views have then been defined :

• The obvious "utterance subtitling" dynamic view D_Utterance
• The ad-hoc transcription view can be used with the utterance< annotation type so as to provide a classical text/video synchronisation
• The static view S_Utterance displays the transcription, temporally linked with the video.

Adding some relations:

• internal_reference inside the discourse.
• adding a rule to the D_Utterance dynamic view taking advantages of these relations by proposing to jump to it.

Designing a S_part_details view that presents the utterances temporally contained in the part of speech it is applied to. For this,

• a containedUtterances query was defined (specialization of the containedAnnotations query, which returns the annotations that are stricly temporally contained in it)
• the S_part_details view itself loops on all the utterances annotations that are returned, and present them using the pre-defined v_enriched_link annotation view. Applying it to an utterance annotation generates HTML code with a link to an annotation, that uses the annotation default view.

Third step: Linking the video to the world (wide web)

Ted Nelson not only makes self references in his speech. He also makes references to the outside web. So as to be able to develop views and usages related to such analyses, we have defined a outsideWeb schema.

We define several annotation types in this schema:

• assistant with name, URL, description, image attributes (we use here the simple application/x_advene-structured annotation MIME-type where each line contains "att_name=att_value")
• operatingSystem with name, description, URL attributes

After annotating the document, we can define several views related to the outsideWeb schema:

• a dynamic view D_assistantsLoop that plays in loop the excerpt of the speech dedicated to assistants (PartOfSpeech annotation with id a3)
• a static view assistants on the fine assistants Ted Nelson talks about, and what they have become. This view is not a generic one, and is directly related to the very video. It uses the D_assistantsLoop to play the corresponding sequences.
• a dynamic view outsideURLs that uses the URL stack to propose links about the concepts at hand
• a dynamic view stopAndBrowse that automatically loads the browser with interesting URL, and pauses the player at the end of the annotation. The BulgariaSpecial rule uses an utterance annotation.

A People annotation type is added to the Summary schema. Keynote speakers appareances in the film are annotated using this annotation type (with two attributes name and URL). The People dynamic view captions the name of the speaker, and proposes web links to personal homepages in the URL stack.

Fourth step: virtual montage

A D_abstract dynamic view has been defined, that consists in designing a fragment trail using the utterance annotation, designing several rules for jumping over non needed parts of the video. We could have done it differently, designing for instance a summaryCuttingRelation in the speech schema, setting up relations between utterances annotations, and designing a summaryCutting dynamic view that uses the relations to automatically jump in the video. Feel free to experiment with this (better) approach.

At the end, the D_Summary view is loaded and the reader is warned of it using an information popup.

The D_laughTour view jumps from on laugh annotation to the following. It uses the followingSameType query.

Fifth step

Designing the general form of the demonstration and the explanations (much more like classical website design)

• the index static view, that presents the package, gives some explanations about views, presents all the views in the package, and gives access to other views. This view was set as default static view for the package (no default dynamic view was set, though it is possible)
• the packageConstruction and the textualTranscription static view
• the doItYourself view

Adding hyperlinks between views.

And now?

You can modify this package. The doItYourself view aims at giving you some directions.

You can also go back to the main index view.