Creating Spatiotemporal Semantic Maps from Web Text Documents - Space-Time Integration in Geography and GIScience - page 156

the term in the article (Lieberman et al. 2007 ; Teitler et al. 2008 ). More recently
TwitterStand automatically imports breaking news from tweets posted by Twitter
users and provides a map interface for reading this news (Sankaranarayanan et al.
2009 ; Jackoway et al. 2011 ).
In the field of GIScience, researchers are growing increasingly interested in
incorporating GIR techniques into their studies. Methods are being investigated
for extracting route directions automatically from text (Klippel et al. 2008 ; Zhang
et al. 2010 ), and research involving oral histories and biographies combine narrative
analysis with a GIS-based time-geographic framework for interactive interpretation,
analysis, and visualization (Kwan and Ding 2008 ). In other recent work, human-
itarian terms are extracted from articles describing the long-term humanitarian
crisis situation in the Sudan and visualized using an earth-based visualization
(Tomaszewski 2008 ; Tomaszewski and MacEachren 2010 ).
Ontologies play an important role in the development of GIR techniques with
regard to providing a knowledge base to improve semantic understanding of text
with respect to searching and extracting appropriate information (e.g., geographic
information). Ontologies deal with the actual nature of the phenomena themselves,
focusing on “the nature and the organization of reality” (Ping and Yong 2009 ,
p. 306). In information science, ontologies have been applied as a tool for knowledge
management and are defined as a concept-based framework using formal methods
that represent entities, attributes, relationships, and values for a specific domain
(Wiegand and Garcia 2007 ; Ping and Yong 2009 ). Well-developed ontologies can
serve as a standard for conceptualizing and understanding domains of interest, and
ontologies support semantic interoperability. For example, associations between dif-
ferent semantics can be achieved by applying ontologies with classification schemes
and hierarchies (Kemp et al. 2007 ). Ontologies can be applied for disambiguation of
geographic names and improving gazetteer interaction (Volz et al. 2007 ;Janowicz
and Kebler 2008 ; Machado et al. 2011 ). Applying ontology in GIR applications can
help to capture natural facts and related information from a human perspective, and
represent extracted text information in a hierarchical form.
In this chapter, a hybrid method of combining gazetteers and ontologies is
presented to model the dynamics of hazard-related events (in this work, tornadoes)
sourced from web texts (Fig. 10.1 ) through spatiotemporal semantic information.
The process can be divided into three main parts: gazetteers and ontology creation,
text processing, and geovisualization. GIR and NLP techniques are used mainly in
the part of text processing. The model aims at providing spatiotemporal semantic
information representation and hidden patterns obtained from text documents.

10.3 Extracting Spatiotemporal Information

from Web Documents

In this section, we employ a framework for modeling movement based on combined
spatiotemporal information extended from web documents and investigate the
role of GIS for representing spatiotemporal details of movement expressed in the