geovisual representations for spatially aware information retrieved from the internet
DESCRIPTION
Geovisual Representations for Spatially Aware Information Retrieved from the Internet. Syed Awase Khirni GIS Division,Dept of Geography University of Zurich. SPIRIT is funded by EU IST Programme Contract Number: IST-2001-35047. Overview. Spatial information retrieval User requirements - PowerPoint PPT PresentationTRANSCRIPT
Geovisual Representations for Spatially Aware Information Retrieved from the Internet
Syed Awase KhirniGIS Division,Dept of Geography
University of Zurich.
SPIRIT is funded by EU IST ProgrammeContract Number: IST-2001-35047
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Overview
Spatial information retrieval User requirements Spatial information seeking process Visualization framework Different ways of geovisualization
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Spatial information retrieval
A large proportion of documents refer to some location on earth.
“Spatially Aware” information retrieval is to provide access to information based not just on thematic, but also on spatial relevance.
Retrieved documents are ranked according to spatial and aspatial characteristics using a variety of algorithms.
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Spatial information retrieval
Task : Representation of documents retrieved from the
internet based on their spatial and contextual aspects.
Constraints : Representation for the internet 2 Dimensional
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
User’s Query
Something Spatial Relationship Somewhere
Keywords nearwithin x kmwithin x min drivein walking distance fromoutside north of ….
Ontology(Place name hierarchy)
(Concepts of interest)/(Points of interest)
HOTELS in CARDIFF
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Query results
A set of geofootprints for each document.
Spatial and contextual ranks
Footprint type: polygon, bounding box, point, line
DOC
Location (geofootprint)
What (contextual info like hotels, railway stations etc.)
History (events like information from news site)
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
User requirements Documents visualized by their spatial context
– ordered by spatial & contextual rank Uncluttered visual representation Topographic/thematic details Geographic concept expansion Description of geographical features Spatial activities they wish to perform at a
particular geographic location – activity maps to support the results retrieved
Based on the user studies carried out to assess SPIRIT user requirements
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Spatial Information Seeking process
Literature Berry-picking model (Bates,1999) Cognitive Styles (Pask & Scott)
Holistic Serialist
User studies Map metaphor Holistic users
Coarser LOD/broader picture Minute details based on their tasks
Serialist users Prefer finest details – region of interest Coarser/finer details based on their tasks
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Framework for Geovisualizations
Results of Query (GML/XML)
DIGESTER Framework
TGN DATA
Visualization Rules (XML)
on the fly
Customized Visualizations(SVG + JS)
process
Deegree Web Map Server
Web Feature Server(SABE DATA /Tele Atlas)
Conformance with spatial data and GIS interoperability standards, OpenGIS GML,SVG W3C,Digital Cartographic Data Standards.
Jess Rules Engine
apache commons lib
GeoTools1.2
gazetteer
CartogramsHolistic
Hyperbolic browserSerialistic
TouchgraphSerialistic
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Cartograms
•Input parameters•Topological neighbourhood• Spatial relevance
Natural History MuseumVictoria and Albert MuseumScience MuseumMadame TussuadsSherlock Holmes Museum
Museums Near London
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Hyperbolic browser & Touch graph
Input parameters. Topological neighbourhood Geometric distance (absolute & relative) Directional details
Visualize spatial and contextual informational hierarchies
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
4
28
7
9
6
5
3
30
13
35
20
25
1134
32
28
18 19
21
12
33
Most Relevant Document
14
31
17
29
26
Geometric distance
23
24
10
16
22
Size of the circle –contextual relevance
Spatialized based on geometric distance &direction
Query: Hotels in London
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Touch graph
Directories
France
World
Regional
EuropeParis
Lyon
MarseilleProvence
Nice
montpellier
NimesSt.Tropez
perpignan
Toulon
Rouen
Rouen
orleans
Reims
HaureTours
limoges
Bordeaux
Toulouse
Perpignan
Limoges
Burgos Bilbao
Gijon
Related Features
Related Resources
+-
Maps & Views
Recreation& Sports
+-
+-Arts & Education +-
St. TropezA Peninsula on the Mediterranean coast of France.
N
Results > 1.BBC Weather centre-Travel
2.GeoPassage:France
3.Focus on France:
4.Just France Vacation Rentals
5.France Geography
cities on the Mediterranean coast of France
Travel & Tourism
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Categorization based on Cognitive Style
Holistic Cartograms
Present a broader overview initially Finer details on selection
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Categorization based on cognitive style
Serialist Hyperbolic browser
Finer details with geometric distances & directions Overview details on user’s selection
Touchgraph Finer details - spatial hierarchy, directions Overview details on user’s selection
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Open Questions
What are the metrics for judging a particular visualization is most suitable based on cognitive styles?
What information is useful for the users to help them visualize their spatial activity and generation of spatial activity maps on the fly?
How do we visualize the spatial characterization of spatial and non-spatial properties which are typical for the data available on the internet?
How do we represent vagueness? (hotels with in 5 min walk from Hauptbahnhof, Zurich)
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Summary
Categorization of visualization based on cognitive styles adopted by the users
Spatial information retrieval User requirements Spatial information seeking process Visualization framework Different ways of geovisualization
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Thanks for Listening!
© 2003 Syed Awase Khirni, GIS Division,Dept of Geography, University of Zurich
Some facts
WWW
Surface webDeep Web
A group that consists of static, publicly available web pages and which is relatively small portion of the entire web.
Size : 4.28 billion pages approx, 50-75 Terabyte of information
Specialized web-accessible databases And dynamic websites which are not widely known by ‘average’ surfers.400 to 550 times larger than the information on the surface web
Size: 550 billion web-connected documents7500 terabytes of high-quality data
78% of all websites are English websites
Indexed 800 million pages(2001)
indexed 2.5 billion individual pages(2002)increasing to 3.1 billion by February 2003.Jan 2004 claimed to cover over 3,307,998,701 pages. Feb 2004, covered ‘6 billion items’: 4.28 billion web pages, 880 million images and 845 million usenet messages
BrightPlanetClaims ‘deep web’ contains550 billion web connected documents
Source: http://www.cyveillance.com/web/downloads/sizing_the_internet.pdf – pg.2