Transcript Good Is Good Enough’ Data Integration

Neogeography: the challenge of channelling
large and ill-behaved data streams
Maurice van Keulen and Rolf de By
LOCATION INTELLIGENCE
FOR SERIOUS APPLICATIONS IN THE LESS DEVELOPED WORLD
 Spatial information is becoming an ordinary commodity
 Google Earth & Maps, MS Bing, NASA’s WorldWind
 Geo-tagging of visited places, meetings, activities; automatic geotagging by personal devices: photo/video camera, cell phone
 Social networks with location intelligence
 In the less developed world, serious applications are slowly becoming
a reality
 Location intelligence for agriculture, health, transportation and
traffic, education, emergency mitigation, electronic payments,
election monitoring, market prices etc.
Kick-off Neogeography
12 Mar 2010
2
SOCIAL NETWORK APPLICATIONS
 Trucking and road availability
 Farming and field suitability
 Traffic and car-pooling
 Emergency response
 Crime and neighbourhood vigilance
 Urban utility monitoring
Kick-off Neogeography
12 Mar 2010
3
NEOGEOGRAPHY
 Neogeography: applications in which geographic information derives
from end-users, not only from official bodies like mapping agencies,
cadastres or other official, (semi-)governmental entities.
 Central problems
 User community is dynamic
 Users contribute information and expect something in return
 Contributed information is not necessarily of good quality or trust
 Contributed information is somewhat unstructured
(contributors cannot be expected to follow strict data schemes and
they may only have access to a cell-phone operated network)
 Need for a new brand of location-based information management
Kick-off Neogeography
12 Mar 2010
4
Example neogeo sites
Importance of neogeography in
disaster response
 In disaster events:
 In situ real-time data
 may be scarce, may be mutually inconsistent, and
 may change over time
 is needed to augment partial knowledge and understanding.
 Communication infrastructure may be damaged.
 All data is welcome, all kinds of data also:





witness reports
photos
audio
videos
human and machine sensor readings
 General public is a powerful information source, and
generally has an incentive to report (911).
The neogeographers in disasters
 People on site
 People affected
 Rescuers and other professionals
 Mobile telephone providers
 Press
 Biggest challenge: how to make sense of large amounts of not very
trustworthy information:
 Can you rely on what unknown sources inform you about?
SYSTEM OBJECTIVE
sms / sensor & satellite data / data from official bodies
Open source
XML-based
spatial data
infrastructure
capable of
orchestrating &
processing
ambiguous/vague
semi/unstructured
geodata
workflows
delivering
personalized
geoservices
XML
geoservices
Kick-off Neogeography
12 Mar 2010
8
SCIENTIFIC CHALLENGES
 Spatiotemporal features
 Extend XML database technology to fully include spatial feature
support (OGC) and support for fully XML-based development of
geoservices and spatiotemporal analysis
 Spatiotemporal vagueness
 Extend information extraction technology to handle ambiguity and
spatiotemporal vagueness in sensor data and explicit natural
language references to the where and when
 Data augmentation and data quality improvement
 Spatiotemporal profiling
 Provide better understanding of user’s information needs by
analyzing historic requests and offered neogeographic data
 User profile pattern matching: finding like-minded users
Kick-off Neogeography
12 Mar 2010
9
CONNECTION WITH OTHER NEOGEO PROJECT
 Space and time issues
 Uncertainty and trust
 Role of the volunteered information
 Difference: handling the map versus handling the data
Kick-off Neogeography
12 Mar 2010
10
THE TEAM
Rolf de By
(ITC)
PhD student @UT
PhD student @ITC
Background: Master @Ain
Background: Master @ITC
Shams University, Cairo
about “Web geoprocessing
about “Automated Arabic
services on GML with a fast
Text Categorization”
XML database”
Strong background in
She proved the feasibility of
natural language processing
some this project’s ideas.
and text/data mining.
Clarisse
Kagoyire (ITC)
Maurice
van Keulen
(UT)
Mena Badieh
Habib (UT)
Jan Flokstra
(UT)
Kick-off Neogeography
12 Mar 2010
11
Think outside the box