Metaphors to Spatialize Information


1. Issues, opportunities and challenges


2. by:


2. 1. Prof. Micha Pazner,


2.2. National Center for Geographic Information and Analysis

2.3. University of California Santa Barbara

2.4. and Department of Geography

2.5. University of Western Ontario, London, Canada

3. Hits as pits on an English language map


3.1. a map that simulates internet search results, depicting "hits" as impact craters

3.2. * HitsAsPits *

4 Islands in a sea of (non) knowledge


4.1. same idea, different metaphor

4.2. * WordIslands *

5. The objectives of this presentation are:


6. The main objective:


6.1. to make the case for professional and systematic use of the field of geographic information science (GISci) and its tools geographic information systems (GIS). 6.1.1. to guide the optimal use of spatial metaphors


6.1.2. for the spatial processing of information


7. A specific objective:


7.1. to introduce a cartographic coordinate system for language


7.1.1. along with the wordfield metaphor


8. The structure of the talk


8.1. The Linguistic Landscape Map and thewordfield metaphor


8.2. A few words on GIS, GlScience, and Spatialization

8.3. Wrap-up: issues, opportunities and challenges


9. The Linguistic Landscape Map and theword field metaphor


9.1. ... for the spatial represention and processing of large amounts of text





10. We would like to create an information space


10.1. into which concepts can be mapped


10.2. and spatial operations can then be performed to


10.2.1. query, model, and visualize relationships between different concepts


11. To this end we would like to create an information space that is compatible with a grid based GIS


11.1. the data model is an aligned stack of map layers


11.2. We would like to be able to put any word or concept in the language in its place on our map


12. The linguistic landscape map metaphor


12.1. we treat the language domain as a map space


12.1.1. initially an empty canvass, our landscape maps will eventually contain abundant information


13. we make use of Cartographic Language (McEachren 1994) to represent Language


13.1. linguistic objects are mapped as landscape elements


14. what's a good landscape element metaphor for a word?


14.1. let's try: word or concept asfield


14.1.1. (a semantic field represented by a mapped field)


15. imagine a rural landscape where a word is a green field


15.1. word as area that has a certain size, shape, perimeter, and internal structure and pattern
15.2.  Word as Field


16. next, we need a system of coordinates


17. How are we to visually represent many concepts next to one another?


17.1. How are we to lay out a word or concept space?


17.2. What will be the planar coordinates of such an information space?


17.2.1. i.e. what will the X and Y axis measure?


18. the Alphabetical Information Space Coordinate System
Alphabetical Coordinate System


18.1. a fixed relative coordinate system for mapping words


19. A word's X coordinate will depend on


19.1. what the first (Capital) letter is, the mix of letters in a word, and word length.


20. The X coordinate for Dog is 361 and for God it is 656.


20.1. Note that while they are the same length and have the same letter mix, even the relative displacement is not identical (off by 5).


21. A word's Y coordinate will also depend on


21.1. what letters are present and in what order, but using a totally different numerical scheme.


22. The Y coordinate for Dog is 844 and for God it is 547.


22.1. Note that while they are the same length and have the same letter mix the result is different, with Dog having the higher value this time.


23. While


23.1. X is a spatial numerical expression of word field size,

23.2. Y is an aspatial numerical expression of word letter pattern.


Word Map24. Words that are less similar in length and composition will have substantially different coordinate values


24.1. Concept is (y=1553 ,x--303)


24.1.1. i.e. "the highest leftmost pixel occupies a grid square whose address is (1553, 303)"


25. However Concept, Concert and Concern would have partial overlap in the current design.


25.1. This can be addressed in iterative refinements of the current addressing system.


25.2. Overlap is a permissible, common, and addressable situation in cartography.


26. What does the distance mean?


26.1. X axis is Alphabetical distance at the global level, with word size distance locally within a letter


26.1.1. note that short words leave space for longer words...


26.1.2. note that the either axis could be designated (i.e. Y could be the Alphabetical axis if there's a good reason...


26.2. Y axis is an arbitrary numerical 'value spreading' function



27. Advantages


27.1. consistent (fixed relative coordinate system)


27.2. dominated by Alphabetical order supported by one, and only one, axis


27.3. lends itself to layering


27.4. lends itself to spatial processing for analysis and visualization


27.5. economical (files are suitable for PC and network traffic)


28. Disadvantages


28.1. not ordered by categories


28.2. not ordered by meanig


28.3. not ordered by context (cf. multidimensional scaling)


28.4. overlap is allowed and must be dealt with using overlay methods


29. A few comments about the use of metaphors


30. Germane metaphors


30.1. weak (not necessarily inferior choices) metaphors versus strong (Couclelis)


30.1 .1. weak: (e.g. "desktop")


30.1.2. weak metaphors that can fade in and out so that we don't become captives of metaphors


30.2. strong (e.g. flight sim)


31. metaphor as design tool


31.1. a good metaphor guides the design of the map in useful directions


31.2. for spatial processing


32. a simple metaphor of a 'boring rural field' is a good place to start


32.1. simple areal feature (can be disjoint)


32.2. we are starting from solid ground...and can build more sophisticate~etaphors from the ground up




33. Summing up our metaphor


34. A cartographic stem-and-leaf plot for text:


34.1. What the graphic stem-and-leaf plot (Tukey) is to numbers,


34.2. the word-as-field map is to text


35. Text built text coordinates


35.1. or concept built concept coordinates


35.2. ie. like data built data measures


36. some attributes are encoded in the X and Y value


36.1. used to create the space (see assumptions below)


37. multifunctioning graphic elements (Tufte)


37.1. the word is explicitly labeled


37.1.1. it is not just a blob- it is readable


38. the shape size and other graphic elements are used as patterns


38.1. by humans


38.2. and can be used by the machine


39. What we have here are words and concepts as areal entities


39.1. These are literally 'semantic fields'


39.2. That are built hierarchically out of pixels


39.2.1. which form letter areas which jointly form the area for a word or concept


40. The Semantic Field metaphor


40.1. Words (text) are used to represent concepts


40.2. However, the words are additionally treated as (graphic) areal entities (duality)


41. The metaphor:


41.1. A word expressing a given concept is viewed as an areal spatial object a field of letters.


42. Assumptions


42.1. we have fi concepts (in the form of words)


42.2. each concept can be mapped to a cartographic location on the information map


42.3. concepts have i attributes (or properties)


42.4. attribute x and attribute y are used to fix the location of the concept in the space


43. attribute z is represented by the numeric value of the pixels that makes up the concept


43.1. attributes Zi (i = 1 to m) are represented using a separate layer for each attribute,


43.2. if for example, the attribute is "3178" for the number of times the word 'concept' was encountered in an Internet search engine query, then the pixels of 'concept' would be coded as 3178.


43.3. note that the z value can represent qualitative nominal values as well as quantitative values.


44. A few words on GIS, GiScience, and Spatialization


45. What essentially are GIS?


45.1. Geographic Information Systems can be represented by a Venn Diagram that has Automated Cartography, Spatial Analysis, and Geographic Database circles.


46. What is all the excitement surrounding GIS about?


46.1 a new spatial literacy


46.1.1. we can not only think spatially (internal representation)


46.1.2. but actually do a lot more about it (external representation)


47. from descriptive to derivative: the ability to progressively generate new information


47.1. --- and the ability to make implicit spatial relations explicit


48. Examples of GISci models that are available


48.1. the cellular exploratory spatial analysis (CESA) subset
48.1.1. * SIP models *

SIP models

48.2. *SIP tasks*
SIP tasks

49. GIS and Geography are strongly intertwined


49.1. something big is happening to Geography...AAG specialty membership as an indicator


49.2. Allied technical fields create an even more massive army:


49.2.1. e.g.: cartography, remote sensing, GPS, math/stats modeling, etc.


49.3. Three Geography GIS textbooks have appeared in January 1997


50. Fundamentals of Geographic Information Systems, Michael N. DeMers, John Wiley & Sons, 1997.


50.l.· "This is a book about geography. It is also about Geographic Information Systems


(GIS)?" (p.3)


· The book emphasizes spatial concepts and ideas over tools.


· Fundamentals ... is aptly titled-the book does an admirable job at laying down in an orderly


fashion the conceptual foundations of a new discipline.


50.3.1. (Pazner 1997, book review to appear in Transactions in GIS, Vol.1 No.3)


51. Spatialization - An important emerging field


51.1. · Spatialization entails the modern use of geographic metaphors and spatial and graphic processing for representing non-spatial information.


51.2. · A conceptually and methodologically challenging fast-growing research field,


51.2.1. - the main problem seems to be how to create the new space


52. A quote:


52.1. "Will there be a Netscape of Virtual Cartography? Or are Netscape, Microsoft, Oracle and every other company creating virtual environments on the Internet already in the cartography business? Schrage will explore the ways technology, complexity, and cartographic expertise may combine to create new marketplaces for maps"


52.1.1. Michael Schrage (MIT), Keynote Speaker on "Multimedia, Hypermaps, and Cartopreneurs:


Mapping the Future of the Marketplace", GIS LIS '96, November, Denver, CO.


53. A growing number of innovative software products are being developed in this area


53.1. Microsoft has hired over 150 people to work in cartography.. 54. Wrap-up: issues, opportunities and challenges


55. General main message: It makes a lot of sense to represent information using spatial metaphors.


55.1. The reason being that substantial benefits can be reaped by applying Geographic Information Science and modern spatial analysis, modeling and visualization.


56. Spatialized information shouldn't be seen just as a candidate for visualization


56.1. but also as a candidate for further processing to 'tease out' derivative and implicit relationships.


57. GISci offers the opportunity to do coupled spatial database querying, analysis, and visualization.


58. Specific main message: to present "woordinates" and word fields.


59. An experimental information space was created that is compatible with a grid based GIS


59.1. Features in this space can benefit from the quantitative and visual functionality of~ spatial processing.


60. "How can we...." questions?


60.1. We have a mechanism for coming up with solutions


60.2. To get to the answer think of it in terms of cartography and GISci and spatial image processing. There is a body of accumulated expertise there.


60.2.1. for example: How do we deal with scale and how can we apply cartographic generalization? Answer: In a similar way that fields can be rescaled and generalized in maps and images. Period.


61 We should think in terms of building multi-disciplinary teams that include GISci specialists, cartographers and geographers.


61.1. and not just linguists, computer scientists, statisticians, psychologists, etc.


62. There is a clear need to develop integrated applications:


62.1. information system (DBMS. query, search, processing)


62.2. geographic information system (e.g. ESRI ArcInfo and ArcJView)


62.3. adapted virtual reality user interface (e.g. MS W95 flight simulator)


62.4. data conversion software between each stage


63. GISci challenges:


64. Creating operational linguistic spatial databases


64.1. E.g. of a whole Dictionary or Thesaurus


65. Using linguistic spatial database functions and models


65.1. Examples of GISci models that are available


65.1.1. the cellular exploratory spatial analysis (CESA) subset *SIPmodels*
65.2. *SIPtasks*


66. Generating output products


66.1. independent


66.2. to be used as input to the next stage


67. The End