My title
 
Chris Culy
AVML, 7 September 2012

My Three Perspectives

1. Language

2. Models

3. Tools

Mention Math Models as inspiration

Visualizations put ideas into our heads …

What questions come to mind?

What makes L/L data different?

1. Language is not mappable

2. Strings are special

3. Individual pieces of L/L data are meaningful

* Some L/L data that is not (strictly) textual at AVML:
 
      Geographic information (e.g. typology, dialect mapping papers)
      Audio/Video (as original data)
      Sound information (waves, spectograms, intonation curves, etc)
      Time duration in Speech/dialog info (e.g. pauses, overlaps, etc)
 
* Connection with original data at AVML: Text Variation Explorer

One visualization, multiple models

Source: http://bl.ocks.org/1341281 by jasondavies

Strings are special

Uses: Structured Parallel Coordinates

Extending the familiar

Uses: xLDD

Connecting with the data

Uses: xLDD

Flexible components

Uses: xLDD

Goethe on seeing

1. Man erblickt nur, was man schon weiß und versteht.
   You glimpse only what you already know and understand.

2. Was man weiß, sieht man erst!*
   You see first what you know!

   Was man weiß, sieht man erst! Denn wie derjenige, der ein kurzes Gesicht hat, einen Gegenstand besser sieht, von dem er sich wieder entfernt, als einen, dem er sich erst nähert, weil ihm das geistige Gesicht nunmehr zu Hülfe kommt, so liegt eigentlich in der Kenntnis die Vollendung des Anschauens.

Sources:
Kanzler F. v. Müller, Unterhaltungen mit Goethe, 24, April 1819, cited in Lexikon Goethe-Zitate
Einleitung in die Propyläen

Perception and Visualization

1. Visual perception = pattern recognition

   

2. "Preattentive" / Tunable features

3. Attention

Sources: VHS Tübingen, photo by L. Lee McIntyre

Testing in new contexts

Uses: ProD

Which visualization for task, user?

Uses: ProD

What next? Grinstein's Grand Challenge

Sources: DoubleTree, Georges Grinstein

Dataset genres

Sources: on archive.org, on archive.org

* Difference between genre and data genre:
 
      data genre is a collection of instances of a genre, with additional properties of the collection  
    e.g.
         an individual letter is an instance of the correspondence genre
         but not of the correspondence-exhange data genre
 
* cf. Discursis workshop/paper, similarities and differences between conversations and correspondence

Continuity visualization 1

Continuity visualizations 2 and 3

* Adjacency High Frequency: EBB uses the high frequency words more than RB does, except for "heart".
 
* Adjacency High Frequency: Connection between word frequency and co-occurrence freq (box sizes)
 
* Adjacency Lower frequency: in this range EBB uses her higher frequeny words than RB does those words
 
* Adjacency Lower frequency: in this range no connection between overall word frequency (or for EBB or RBB) and co-occurrence freq
 

 
* Hopefully, they are asking questions: What about X? Could you show Y? Why not Z?

Visualizations put ideas into our heads!

Thank You

christopher.culy@uni-tuebingen.de

http://www.sfs.uni-tuebingen.de/~cculy/