También disponible en Español


The digital magazine of InfoVis.net

The Landscape Metaphor
by Juan C. Dürsteler [message nº 168]

The landscape metaphor is one of the most interesting and used one. We describe here its main features and a set of examples of how are they used.
Newsmaps19990417.jpg (46119 bytes)
Newsmaps. Image of the news of the 17th of April, 1999 through the defunct newsmaps.com website. Created using the ThemeScape technology.
Source: The author's archive. 
Click on the image to enlarge it. 

In the previous message we spoke about ThemeScape, a visualisation that uses the landscape metaphor to represent in the form of a topographical map a document space where distance between each two documents is inversely proportional to their similarity.

There is considerable evidence that the human brain uses what in cognitive psicholgy are called "schemata". Although the rigurous definition of a schema would require more space than the one we have here available, we can consider it as a structured set of generic knowledge that can be applied to many specific situations. 

They are made of variables, with their corresponding values, and relationships between them. A scheme can be "hit". For example Robert hit the ball or the car hit the walkway. The schema for "hit" is applied to both cases and summarises a whole set of knowledge and experiences related to that action.

When designing a visualisation, the designer uses his or her set of mental schemata to elaborate a visual metaphor (see number 91) that will be implemented into the visualisation. If the mental schemata of the receptor of the visualisation do not match in some way to those of the designer it's more than possible that the receptor won't understand the visualisation or will not be able to extract all of its possibilities.

A clear example are traffic signals. Whoever that has followed a driving course is able to catch the meaning of a traffic signal in a fraction of a second. Nevertheless if we show the same signal to someone that hasn't been taught how to interpret it or has never seen one, it will be very difficult for him or her to get the meaning in its complete magnitude, even if staring for a long time in front of it trying to decipher it.

All this digression has as its subject the landscape (or spatial) metaphor, that has been essayed in different forms during the last years. It appears that the concept of [topographic] map and the mental schemata associated to it are very widespread. Maybe this is due to the ubiquity of maps, known since very ancient times and used throughout many centuries. According to others it's because they constitute a natural mataphor for the human beings that have evolved in a spatial environment that required a knowledge of terrain and space in order to survive. 

In the end, it seems that decoding a map, be it spatial or thematic, results quite natural for most of the human beings, requiring just very little training. This doesn't appear to be the case with more recent developments of more abstract nature, like parallel coordinates or hyperbolic geometry, just to mention a couple of examples.

Beyond ThemeScape we can identify other interesting initiatives that use this visual metaphor in one way or the other, like (among others)


SpacecastSurf.gif (42935 bytes) CartogramaBushKerry.gif (60049 bytes)
Spacecast: One of the thematic visualisations, in a perspective view, that helps to enhance the differentiation of the importance of some topics over the others.
Source: Image as can be seen in the Spacecast website. 
Click on the image to enlarge it.
Cartogram of the 2004 U.S. presidential elections.  Prepared by Sara I. Fabrikant, it depicts the percentage of vote (the redder the more votes for Bush, the bluer the more for Kerry) down to the county level which area is proportional to the amount of votes emitted.  This leads to a distortion of the map 
Source: Image as can be seen in Sara I. Fabrikant website. 
Click on the image to enlarge it.

Spacecast is a three year project, led by Sara I. Fabrikant whose goal is to create an experimental test bed for the graphical representation of very large database archives based on spatial metaphors (spatializations). During the first year (already gone) thay have studied a wide range of spatial metaphors. The second year has been devoted to experiment with different ways to work with them, while the third year, now running, will be dedicated to the work with 3D geospatial metaphors.


Anyone browsing an Atlas has seen maps where, on top a geographical basis, there are economical, social or statistical magnitudes overlaid. Cartograms are  simplified maps, usually distorted, that allow the user to study a particular phenomenon. Very interesting are the different cartograms, also elaborated by Sara I. Fabrikant, devoted to the results of 2004 US. presidential elections.


VxInsight.gif (108574 bytes) VxInsight2.gif (104868 bytes)
VxInsight: A view of the topics depicted in the form of mountains and valleys.
Source: Image as can be seen in VxInsight website. 
Click on the image to enlarge it.
VxInsight: Another view. Height depends here on the density of objects (gray dots)
Source: Image as can be seen in VxInsight website. 
Click on the image to enlarge it.

VxInsight  created (and patented US 5,987,470) by the Sandia National Laboratory in Albuquerque, NM, USA. uses the spatial metaphor to represent in a conceptually similar way to ThemeScape a set of documents, where a measure of similarity is defined so that you can compute a "distance" between any two documents, in a way where the coordinates assigned to each document make that two similar documents lay close one another, at a short distance, while dissimilar ones get spatial coordinates that lie apart, far away, in the representation plane. 

Document density is coded as height, so that thematic zones with a high number of documents appear as mountains, while sparse zones appear as plains.


WebSOMZoom.gif (111905 bytes) WebSOM.gif (21145 bytes)
WebSOM: Representation of the contents of Usenet's  newsgroups by means of a Self Organising Map (SOM) 
Source: Image as can be seen in WebSOM website. 
Click on the image to enlarge it.
WebSOM: A zoom of a part of the previous view. Each white dot is a document that you can consult just by clicking on it.
Source: Image as can be seen in WebSOM website. 
Click on the image to enlarge it.

WebSOM is a project of Helsinki University of Technology (HUT) led by Teuvo Kohonen that uses a similar metaphor computed with a very different algorithm. Here close up documents are also similar, but a neural network algorithm is used to create a SOM or Self Organising Map. This type of map were reviewed already in number 39 of this e-zine.

 Islands of Music

IslandsOfMusic.gif (35084 bytes) IslandsOfMusicEtiquetado.gif (35789 bytes)
Islands of Music: Map of the "islands" representing musical genres. The analysis of MP3 files allows the system to automatically classify them, finding also their (perceived) psycho-acoustic similarity. 
Source: Image as can be seen in Islands of Music website 
Click on the image to enlarge it.
Islands of Music: Another view, with labels that refer to some musical properties of the genres they represent.  
Source: Image as can be seen in Islands of Music website 
Click on the image to enlarge it.

Islands of Music is the outcome of the PhD thesis of Elias Pampalk. Again you get a map, in this case of islands surrounded by an ocean, of documents that here are musical pieces. These documents contain only the piece itself in MP3 and the musical genre they belong to. Once more, distance is equivalent to similarity. Mountains and valleys correspond to musical genres.

The originality of Islands of Music lies in the fact that similarity is computed taking only into account the contents of the MP3 file and a psycho-acoustic approximation that computes the similarity perceived when listening two given musical pieces. With this you can build up a map of musical genres that are labeled with words describing certain rhythmic and musical properties of the pieces. 

The landscape metaphor, and spatialisations in general, constitute very attractive ways of representing big amounts of data in an intuitive way. The notions of distance and height are easily understood by many people, even when they encode other variables like similarity, density, or prevalence of a certain term. Let me put a nostalgic note if I say that the old newsmaps is still one of my preferred visual metaphors.

Links of this issue:

http://www.infovis.net/printMag.php?num=167&lang=2   Num 167 Patent Analysis
http://www.infovis.net/printFicha.php?rec=revista&num=91&lang=2   Num 91 Visual Metaphors
http://www.geog.ucsb.edu/~sara/html/research/spacecast/spacecast.html   Spacecast website
http://www.geog.ucsb.edu/~sara/html/mapping/election/election04/election.html   2004 US presidential election (cartograms)
http://www.cs.sandia.gov/projects/VxInsight.html   VxInsight website
http://websom.hut.fi/websom/milliondemo/html/root.html   WebSOM demo
http://websom.hut.fi/websom/   WebSOM website
http://www.hut.fi/   Helsinki University of Technology (HUT)
http://www.infovis.net/printMag.php?num=39&lang=2   Num 39 Mapas de Kohonen
http://www.oefai.at/~elias/music/index.html   Islands of Music website
© Copyright InfoVis.net 2000-2014