Geography Reference
In-Depth Information
same 126 processed spatiotemporal references, the system performed with 108
correct references, 18 incorrect references, and 15 missed spatiotemporal references.
Based on this performance, precision and recall are calculated as 0.81 and 0.89
respectively. For this evaluation, for both the human evaluators and the system, it
appears that as the total number of processed spatiotemporal references increases,
the values of precision and recall decrease. Further testing is necessary to confirm
the patterns observed here, but so far it appears the methods can lead to appropriate
spatiotemporal assignments.
10.7
Conclusions
In this research, GIR methods and natural language processing techniques are
combined such that spatiotemporal information is extracted automatically from sets
of web documents to associate the semantics of events with spatial and temporal
information. This offers a method to track the evolution of events over space-
time described in web text and present this information in a geovisualization.
A framework is presented for spatiotemporal semantic information extraction that
is based on gazetteer and ontology creation, text processing, and geovisualization.
Rules have been developed for combining extracted spatial and temporal terms in
order to represent hazard events or other moving phenomena through space-time as
they are described in text. Spatial and temporal gazetteers have been developed to
aid the information retrieval process. These techniques are further extended through
the use of ontologies that allow the semantics of events to be associated with spatial
and temporal information. A hazard ontology and semantic gazetteer is invoked
to support semantic representation of extracted events sourced from web texts.
Including semantics contributes to a better understanding of spatiotemporal patterns
of events from both natural and human perspectives. Using ontologies represents
extracted text information in a hierarchical structure, and affords the opportunity
to represent semantics at multiple granularities, giving further insights about events
from a spatiotemporal perspective.
References
Chen, X., & Kwan, M.-P. (2012). Choice set formation with multiple flexible activities under space-
time constrains. International Journal of Geographic Information Science, 26 (5), 941-961.
Demsar, U., & Virrantaus, K. (2010). Space-time density of trajectories: Exploring spatio-temporal
patterns in movement data. International Journal of Geographic Information Science, 24 (10),
1527-1542.
Dodge, S., Weibel, R., & Lautenschutz, A.-K. (2008). Taking a systematic look at movement:
Developing a taxonomy of movement patterns. Information Visualization, 7 (3), 240-252.
doi: 10.1057/palgrave.ivs.9500182 .
Search WWH ::




Custom Search