Creating and Using Geospatial Ontology Time Series in a Semantic Cultural Heritage Portal

Abstract

Content annotations in semantic cultural heritage portals commonly make spatiotemporal references to historical regions and places using names whose meanings are different in different times. For example, historical administrational regions such as countries, municipalities, and cities have been renamed, merged together, split into parts, and annexed or moved to and from other regions. Even if the names of the regions remain the same (e.g., "Germany"), the underlying regions and their relationships to other regions may change (e.g., the regional borders of "Germany" at different times). As a result, representing and finding the right ontological meanings for historical geographical names on the semantic web creates severe problems both when annotating contents and during information retrieval. This paper presents a model for representing the meaning of changing geospatial resources. Our aim is to enable precise annotation with temporal geospatial resources and to enable semantic search and browsing using related names from other historical time periods. A simple model and metadata schema is presented for representing and maintaining geospatial changes from which an explicit time series of temporal part-of ontologies can be created automatically. The model has been applied successfully to represent the complete change history of municipalities in Finland during 1865-2007. The resulting ontology time series is used in the semantic cultural heritage portal CULTURESAMPO to support faceted semantic search of contents and to visualize historical regions on overlaying maps originating from different historical eras.

Full text

Creating and Using Geospatial Ontology Time Series in
a Semantic Cultural Heritage Portal
Tomi Kauppinen1, Jari Väätäinen2, and Eero Hyvönen1
1Semantic Computing Research Group (SeCo)
Helsinki University of Technology
and University of Helsinki, Finland
http://www.seco.tkk.fi/
firstname.lastname@tkk.fi
2Geological Survey of Finland
http://www.gtk.fi
firstname.lastname@gtk.fi
Abstract. Content annotations in semantic cultural heritage portals commonly
make spatiotemporal references to historical regions and places using names
whose meanings are different in different times. For example, historical adminis-
trational regions such as countries, municipalities, and cities have been renamed,
merged together, split into parts, and annexed or moved to and from other regions.
Even if the names of the regions remain the same (e.g., “Germany”), the underly-
ing regions and their relationships to other regions may change (e.g., the regional
borders of “Germany” at different times). As a result, representing and finding
the right ontological meanings for historical geographical names on the semantic
web creates severe problems both when annotating contents and during informa-
tion retrieval. This paper presents a model for representing the meaning of chang-
ing geospatial resources. Our aim is to enable precise annotation with temporal
geospatial resources and to enable semantic search and browsing using related
names from other historical time periods. A simple model and metadata schema
is presented for representing and maintaining geospatial changes from which an
explicit time series of temporal part-of ontologies can be created automatically.
The model has been applied successfully to represent the complete change his-
tory of municipalities in Finland during 1865–2007. The resulting ontology time
series is used in the semantic cultural heritage portal CULTURESAMPO to sup-
port faceted semantic search of contents and to visualize historical regions on
overlaying maps originating from different historical eras.
1 Introduction
Geospatial ontologies define classes and individuals for representing e.g. geographic
regions, their properties, and mutual relationships. By sharing ontological resources in
different collections and application domains, interoperability in terms of geographical
locations can be obtained, and intelligent end-user services such as semantic search,
browsing, and visualization be facilitated [9, 24,23]. For example, in the semantic por-

2
tal MUSEUMFINLAND3[10] a location partonomy4was used for annotating museum
artifacts with metadata about the place of manufacture and the place of usage.
A lesson learned during this work was that geography changes rapidly, which makes
it hard 1) to the content annotator to make correct references to spatiotemporal regions
and 2) to the end-user to understand the changes in historicalgeography and, as a result,
to formulate the queries. For example, many artifacts in MUSEUMFINLAND originate
from regions that no longer exist and/or have not been a part of Finland but of Russia
after the Second World War. Finding the right names for querying, understanding to
which regions on the map the names refer to at different times, and understanding how
old historical names relate to modern Finnish and Russian geography creates, at the
same time, both a semantic challenge for the technology and an important part of useful
content to learn when using the portal.
This paper addresses two essential needs from the end-user point of view when
using historical geographic regions in a cultural heritage portal:
Ontology-based spatiotemporal search It is necessary to be able to use both histor-
ical and modern regions as search concepts e.g. in a view-based, or multi-facet
search [7, 20, 18,6]. The idea is that regions offer one view to a content and they
can be used to select a subset of the content by specifying constraints. For exam-
ple, selecting “Finland (1945-)” from a facet view would refer to a part of Europe
relating to the modern post-war Finland.
Visualization of concepts It is necessary for the end-user to be able to see where the
historical regions are on the map in a proper temporal context. Moreover, there
should be a way of visualizing the spatial relationship between the old regions and
the modern ones on the maps in order to relate history with the world of today. Cre-
ation of several layers of information on maps is a common way to visualize maps
[3] and related content at the same time. In our case, we decided to apply this idea
to overlaying historical and modern maps to visualize spatiotemporal relationships
of regions, and to display related cultural content on the maps. Such map visual-
izations also help in finding the right concepts for a search and for presenting the
search results.
To successfully meet the above needs the following requirements can be set for the
ontology creation and management:
1. Concepts representing the regions from different time intervals need to be identified
by URIs, put into a valid spatial part-of hierarchy, and mapped with each other in
the temporal dimension.
2. Essential geographical properties, such as coordinates of points or polygonal bound-
aries, time span, size, and names of the historical regions, need to be assigned to
the URIs of the regions.
3http://www.museosuomi.fi
4This partonomy is a part-of hierarchy of individuals of the classes Continent, Country, County,
City, Village, Farm, etc.

3
To meet these requirements, it is essential that a geospatial ontology used in a se-
mantic cultural system can represent change in time [15]. In current historical geo-
vocabularies and ontologies, such as the Getty Thesaurus of Geographic Names (TGN)5,
historical regions may be found, but the aspect of change is usually missing. For exam-
ple, in the TGN the historical city of “Rome” in Italy has an entry as an inhabited place,
but its development from an Etruscan city of the 8th century BC to its declination in
330 AD is described only as a piece of literal text.
In this paper, we present a simple metadata schema and a model for representing
geospatial changes and for maintaining them as an RDF repository. A method for con-
structing a time series of geospatial, temporal ontologies (an ontology time series) from
the filled metadata schema is discussed, and a reasoning mechanism to infer properties
(size), relationships, and mappings between spatiotemporal regions is then presented.
To test and evaluate the approach, the system was used in a case study of creating a
complete model of the changes of the Finnish municipalities in 1865–2007. Finally,
we present how the resulting ontology time series has been applied to creating intel-
ligent services and map-based visualizations in the semantic cultural heritage portal
“CULTURESAMPO—Finnish Culture on the Semantic Web”6[12].
2 Modeling Geospatial Changes
2.1 Analysis of Change Types
We analyzed the kinds of regional changes of municipalities in Finland7between years
1865 and 2007. Table 1 lists the change types found and their quantities.
Change type Quantity
Establishment (A region is established) 508
Merge (Several regions are merged into one) 144
Split (A region is split to several regions) 94
Namechange (A region changes its name) 33
Changepartof (Annexed (to a different country)) 66
Changepartof (Annexed (from a different country)) 1
Changepartof (Region moved to another city or mu-
nicipality) 256
Total sum 1102
Table 1. Different types of regional changes between 1865 and 2007 in Finland.
An example of a merge is depicted in Figure 1. In the year 1922 Nummi-Pusula
was formed via the unification of two former municipalities, Nummi and Pusula. This
5http://www.getty.edu/research/tools/vocabulary/tgn/
6http://www.kulttuurisampo.fi
7As collected by the Geological Survey of Finland

4
means that the old notions of Nummi and Pusula became obsolete after 1922, and the
new concept of Nummi-Pusula was introduced.
In Figure 2, there is an example of a split. Pirkkala was split into two municipalities,
Pohjois-Pirkkala and Etelä-Pirkkala in 1922. In Figure 3 there are two examples of
name changes. In year 1938 Pohjois-Pirkkala was renamed Nokia. At the same time, its
neighbor Etelä-Pirkkala was renamed to Pirkkala. Finally,in Figures 4,5 and 6 there are
three different examples of changes in a partition hierarchy. Figure 4 depicts a change
where Petsamo was annexed fromRussia to Finland in 1920. Figure 5 depicts a change
where Petsamo was annexed from Finland to USSR in year 1944. Finally, Figure 6
depicts a different change: Vuosaari was moved from Helsingin maalaiskunta to the
city of Helsinki in year 1966.
These changes always change not only the sizes of the regions in question but also
the partition hierarchy. This means that from year 18658until 2007 there are 142 dif-
ferent kind of partition hierarchies of historical "Finlands".
Fig.1. An example of a merge. Fig.2. An example of a split.
Fig.3. Two examples of name changes. Fig. 4. An example of a change where a part
of a region is moved.
Modeling all these different 142 temporal partition hierarchies of Finland, the re-
sources and their mutual relationships, as separate ontologies by hand would be hard.
Instead, we propose utilization of a simple schema for representing changes, and using
an automated process for generating the different partition hierarchies.
2.2 A Schema for Representing and Maintaining Changes
The change types of Table 2 can be represented in terms of a few metadata fields (el-
ements) listed in the Metadata Schema of Changes (see Table 2). An other metadata
8In year 1865 first municipalities were established in Finland.

5
Fig.5. Another example of a change in a
partition hierarchy. Fig. 6. A third example of a change in a par-
tition hierarchy.
schema, the Metadata Schema of Current Places is meant for maintaining the contem-
porary places, like cities, municipalities and countries (see Table 3) and the Metadata
Schema of Historical Places for properties such as boundaries of historical regions (see
Table 4). Note that the last two schemas are very similar and could also be integrated.
Different fields of the Metadata Schema of Changes, such as place,date,change
type and from and to-fields are filled up with the changes and resources they concern.
For example, a change concerning the annexing of Petsamo from Finland to USSR on
1944-09-19 has an own instance conforming to the metadata schema, with the corre-
sponding fields filled up (from=Finland,to=USSR,movedpart=Petsamo,date=1944-
09-19, and so on). Notice that for each region modified by a change, a separate instance
conforming to the metadata schema is created.
Field Definition C Value range
identifier Identifier for a change 1 Change ontology
(automatically generated)
date The date of the change 1 W3CDTF (ISO8601)9
place Place field 1 Location ontology
place type The type of the place 1 Location ontology
change type The type of the change 1 Change type ontology
(either establishment, merge,
split, namechange, or
changepart, (see Fig. 1–6)
from From where there are areas 1..* Location ontology
moving to in the change
to To where there are areas moving 1..* Location ontology
to in the change
movedpart Which part(s) are moving in the change 0..* Location ontology
(if they can be named)
Note: Only used for changepartof
description Description of a change 1 String
Table 2. The Metadata Schema of Changes. Cardinalities are presented in the column C.
9For an implementation as an XML Schema Date, see http://www.w3.org/TR/xmlschema-2/

6
Field Definition C Value range
identifier Identifier for a place 1 Location ontology
place name Current place 10. 1 Location ontology
size Size of the place in square kilometers 0..1 Double
partof Which other administrational region 1 Location ontology
this region is a part of
point Representative point of the place 0..* ISO 19107, WGS84
polygonal boundaries Polygonal boundaries of the place 0..* ISO 19107, WGS84
maps Maps of the place 0..* Map ontology
Table 3. The Metadata Schema of Current Places.
Field Definition C Value range
identifier Identifier for a place 1 Location ontology
place name Place name 1..* Location ontology
size Size of the place in square kilometers 0..1 Double
partof Which other administrational 0..1 Location ontology
region this region is part of
point Representative point of the place 0..* ISO 19107, WGS84
polygonal boundaries Polygonal boundaries of the place 0..* ISO 19107, WGS84
measurement date Date when the properties were valid 1 W3CDTF (ISO8601)
(e.g. boundaries)
maps Maps of the place 0..* Map ontology
Table 4. The Metadata Schema of Historical Places.

7
3 Creating an Ontology Time Series and Overlap Mapping
The previous section described three schemas used for creating an ontology time series.
An ontology time series [15] defines a set of geospatial ontologies, including partonomy
hierarchies for different time spans. This knowledge is represented in terms of RDF
triples [1], where a resource (subject) is characterized by an identity (URI) and related
property (predicate) values (object) in the form <sub ject,predicate,ob ject >.
The following example motivates the creation of different temporal ontologies. Let
us assume two RDF triples represented in a table, utilizing the name space dcterms of
Dublic Core Metadata Terms 11 and another namespace location:
Subject Predicate Object
location:Monrepos dcterms:isPartOf location:Vyborg
location:Vyborg dcterms:isPartOf location:Russia
These triples could come from an RDF repository containing a traditional parton-
omy hierarchy that define the fact that the famous park Monrepos is a part of the city
called Vyborg which in turn is a part of Russia. This is true for the ontology of the year
2007. However, the two RDF triples
Subject Predicate Object
location:Monrepos dcterms:isPartOf location:Viipuri
location:Viipuri dcterms:isPartOf location:Finland
define the historical fact that Monrepos is a part of Finland—this was true in 1921–
1944. As we can see, these two sets of RDF triples would confuse a reasoner and the
end-user, because location :Monrepos would be a part of two non-intersecting regions
location :Russia and location :Finland (assuming that dcterms:isPartOf is transitive).
To overcome this problem, our ontology time series is populated with different tem-
poral parts of places which are described by a metadata schema. Examples of temporal
parts of location:Viipurin mlk are location:Viipurin mlk (1869-1905),location:Viipurin
mlk (1906-1920),location:Viipurin mlk (1921-1944) and location:Vyborg(1944-). All
these temporal parts have different polygonal boundaries, different sizes, and some of
them are also in a different partonomy hierarchy.The ontology population process pro-
ceeds in the following way.
First, a place is created in the RDF repository (like location:Viipurin mlk). Based on
the two sequential changes in the Metadata Schema of Changes for Viipurin mlk, that
happened e.g. in 1906 and in 1921, a temporal part location:Viipurin mlk (1906-1920)
is created and added to the union of 12 location:Viipurin mlk. Similarly, by examining
the next two sequential changes concerning the place, additional temporal parts (like
location:Viipurin mlk (1921-1944)) are created. If there are no more changes for that
place, then the place has ceased to exist (like location:USSR (1944-1991)) or it is a
10 Present in year 2007 as of writing this paper.
11 http://dublincore.org/documents/dcmi-terms/
12 owl:unionOf is used, http://www.w3.org/TR/owl-guide/

8
contemporary one (like location:Helsinki (1966-)). Whether the place is a contemporary
one is checked from the Metadata Schema of Current Places.
Second, the properties for the temporal parts of places are retrieved from the Meta-
data Schema of Current Places and from the Metadata Schema of Historical Places,
depending whether the place in question is an existing one or has ceased to exist. This
phase creates RDF triples representing, for example, the polygonal boundaries, the cen-
ter point, the size, and partonomical relationships of the temporal part of the place. For
example, two different partonomy hierarchies of our previous example of Monrepos is
defined by four triplets
Subject Predicate Object
location:Monrepos(1921-1944) dcterms:isPartOf location:Viipuri(1921-1944)
location:Viipuri(1921-1944) dcterms:isPartOf location:Finland(1921-1944)
location:Monrepos(1991-) dcterms:isPartOf location:Vyborg(1991-)
location:Vyborg(1991-) dcterms:isPartOf location:Russia(1991-)
In addition, there are triples defining that different temporal parts of Monrepos be-
long to the same union of location:Monrepos, and triples defining different properties
for temporal parts.
A temporal ontology [15] includes all temporal parts (of places) of some time span.
For example, a temporal ontology of the year 1926 would include location:Viipurin
mlk (1921-1944) because the year 1926 is within the range 1921–1944. Furthermore,
the ontology contains all the partonomical relationships of those temporal parts that are
valid during its time span.
Next, when all the places, their temporal parts and properties are created in the on-
tology time series, a model of changes is created based on the fields of the Metadata
Schema of Changes. In each change there is something before the change (like loca-
tion:Viipurin mlk (1869-1905)) and something after the change (like location:Viipurin
mlk (1906-1920)). This is expressed with properties before and after. In practice, the
following types of RDF triples are added to the repository:
Subject Predicate Object
change:change15 change:before location:Viipurin mlk(1869-1905)
change:change15 change:after location:Viipurin mlk(1906-1920)
change:change15 change:after location:Nuijamaa(1906-1944)
change:change15 change:date ”1906-01-01”
change:change15 change:changetype change:split
These triples are used as an input for a previously published method [14, 15] to
create a global overlap table between different temporal parts in the ontology time
series. This table tells how much each place overlaps with the others. The repository is
filled by following kind of triples based on the global overlap table calculation:
Subject Predicate Object
overlapping:overlap31 overlapping:overlaps 1.0
overlapping:overlap31 overlapping:overlappedBy 0.3131
overlapping:overlap31 overlapping:argument1 location:Viipurin mlk(1869-1905)
overlapping:overlap31 overlapping:argument2 location:Nuijamaa(1906-1944)

9
For example, since the size of Nuijamaa (1906-1944) is 407 square kilometers and
the size of Viipurin mlk (1869-1905) is 1300 square kilometers, Nuijamaa overlaps
Viipurin mlk by value 407/1300 =0.3131 and is overlappedBy by Viipurin mlk by
value 407/407 =1.0 after the split (cf. the example above).
The Figure 7 illustrates the global overlap table by depicting overlaps with colors
between a selected set of regions. The black color indicates a full 100% overlap between
the temporal parts and the white color a 0% overlapping, accordingly.Different shades
of grey indicate the level of overlapping: the darker the box, the greater is the overlap.
From this illustration it is easy to see the mutual asymmetric overlaps between the
temporal parts, and that the overlap-relation in this case is fairly complicated.
Fig. 7. Overlaps between temporal parts of places visualized using colored boxes. The black color
indicates a full 100% overlap between the temporal regions and the white color a 0% overlap,
accordingly. Different shades of grey indicate the level of overlap between regions: the darker the
box, the greater is the overlap between the regions.

10
Fig.8. A set of changes collected as a spreadsheet table.
4 Creation of a Finnish Spatio-temporal Ontology
The metadata schemas and methods described in the previous sections were imple-
mented to create a Finnish Spatio-temporal Ontology, an ontology time series of Finnish
municipalities over the time interval 1865–2007.
The metadata schemas were implemented as a spreadsheet tables13 for easy editing.
Figure 8 shows a screenshot of the Metadata Schema of Changes. Different schema
fields, such as place,date,change type,from, and to, are filled up with resources and
values. For example, the split of Viipurin mlk (1869-1905) into Nuijamaa (1906-1944)
and Viipurin mlk (1906-1920) is seen on the row 1197, and the annexing of Viipuri
from Finland to USSR on 1944-09-19 is on the row 1195. Most changes have a natural
language explanation of the event.
The methods for creating an ontology time series from the metadata schemas were
implemented using Java and Jena Semantic Web Framework14 [2]. The resulting RDF
repository contains 1105 different changes and 976 different temporal parts of 616 dif-
ferent historical and modern places, meaning each place has on average 1.58 temporal
parts. For example, the place resource location:Viipurin mlk got the temporal parts loca-
tion:Viipurin mlk (1869-1905),location:Viipurin mlk (1906-1920),Viipurin mlk (1921-
1943), and location:Viipurin mlk (1944-). The temporal parts and their partonomy hier-
archies in the RDF repository constitute 142 different temporal ontologies between the
years 1865 and 2007, each of which is a valid model of the country during its own time
span.
5 Applications for Spatiotemporal Search and Visualization
Two case applications were created to utilize the resulting ontology time series in real
application scenarios. The first one uses partition hierarchies of different time spans in
13 We used the freely available OpenOffice Calc (http://www.openoffice.org/)
14 http://jena.sourceforge.net/

11
faceted search facilitating ontology-based spatio-temporal search. Both historical and
modern regions can be used as search categories. To illustrate this, in Figure 9 two cat-
egories corresponding to temporal parts of location:Helsinki, namely location:Helsinki
(1966-) and location:Helsinki (1640-1946) are selected in a search facet, and related
items from cultural collections are retrieved.
This functionality is included in the semantic CULTURESAMPO portal [12] that cur-
rently contains over 32 000 distinct cultural objects. The annotations of the objects were
enriched automatically by comparing the time span and place of each annotation with
those of the temporal parts of places. If they overlapped and place names matched, then
the annotation was enriched accordingly. CULTURESAMPO also allows for searching
with places on a map as illustrated in Figure 11. By clicking a place on a map, the items
annotated with that place are retrieved and shown on the right side of the map. Further-
more, the user can formulate a search query as a polygon by pointing out npoints on a
map. All the places that have a point inside that polygon are retrieved and the content
related to those places are listed on the right side of the page.
Fig.9. Temporal parts of places used as a
search constraint in CultureSampo.
Fig.10. Using multiple maps simultane-
ously. A historical Karelian map depict-
ing the city of Viipuri is shown semi-
transparently on top of a modern satellite
image provided by the Google Maps ser-
vice. Temporal parts of places on the left
can be used to select different maps. The
search for cultural artefacts can be con-
strained in this view by pointing our n
points on a map.
Our second application [13] utilizes the ontology time series in visualizing historical
and modern regions on top of maps and satellite images. This answers to the need for
visualizing spatiotemporal places: it is necessary for the end-user to be able see where
the historical regions are on the map in a proper temporal context. Figure 12 illustrates
the application. Historical regions, i.e. temporal parts of places, can be selected from
a drop-down menu on the left. Here a temporal part location:Viipuri(1920-1944) of
location:Viipuri is selected. As a result, the polygonal boundaries of Viipuri (1920–

12
1944) are visualized on a contemporary Google Maps satellite image, map, or on a
historical map. In addition, modern places from ONKI-Geo [11] that are inside the
polygonal boundaries of the historical region are retrieved in a mash-up fashion, and
can be used to browse the map. The content related to location:Viipuri(1920-1944)
is listed in this view on the right. Furthermore, content from historical regions that
overlap location:Viipuri(1920-1944) are listed as recommendations. The overlappings
are looked up from the global overlap table.
Historical maps can be shown on top of the contemporary maps, as depicted in Fig-
ure 10. In the middle, a contemporary satellite Google Maps image of the city of Viipuri
in the Karelia region is shown. In the middle, a smaller rectangular area is shown with a
semi-transparent15 old Karelian map that is positioned correctly and is of the same scale
as the Google Maps image. This smaller view shows the old Viipuri, an old Finnish city
that nowadays is a part of Russia. The place cannot be found in current maps as it was,
which makes it difficult for modern users to locate the place geographically. In order
to move around the user is able to use the zooming and navigation functions of Google
Maps and the historical view is automatically scaled and positioned accordingly.
To provide the historical maps, we used a set of old Finnish maps from the early
20th century covering the area of the annexed Karelia region before the World War II.
The maps were digitized and provided by the National Land Survey of Finland16. In
addition, a geological map of the Espoo City region in 1909, provided by the Geolog-
ical Survey of Finland17, was used. This application is also included in the CULTURE-
SAMPO portal [12].
Fig.11. A search with regions without tem-
poral extensions.
Fig. 12. Temporal parts are used to visualize
polygonal boundaries of historical regions
in CULTURESAMPO and for searching his-
torical artifacts.
15 We use transparency libraries provided by http://www.kokogiak.com/ which allow the alter-
ation of the level of transparency.
16 http://www.maanmittauslaitos.fi/default.asp?site=3
17 http://en.gtk.fi

13
6 Conclusions
6.1 Contributions
This paper presented an analysis of change types in historic regions, a model of changes
based on the analysis and an ontology time series from the model, and a practical tool
for maintaining the RDF repository of changes. We have also succesfully applied an
existing method [15] to create a global overlap table from the repository of changes.
We have evaluated the usability of the resulting ontological structure—the ontology
time series—in two real life applications for information retrieval and for visualization
in a semantic cultural heritage portal.
These applications can be used for teaching where historic regions have been and
how they are related with each other in a partonomy hierarchy. The visualization is
made using a rich set of historic maps, modern maps, satellite images, and polygonal
boundaries. In addition, the applications can be used for retrieving historical cultural
content related to the regions. The relationship is explicated for the user indicating
whether the content has been found, used, manufactured, or located in a specific region.
Old maps and names on them could be of substantial benefit when using visualiza-
tion in annotating or searching content in cultural heritage systems. The idea of using
overlaid transparent maps is useful when comparing geo-information from different
eras (e.g., how construction of cities has evolved) or from different thematic perspec-
tives (e.g., viewing a geological map on top of a satellite image). We believe that map-
based views of historic locations together with rich, precisely, and spatio-temporally
annotated cultural content offer a good use case of semantic web technologies for solv-
ing real life interoperability and information retrieval problems.
6.2 Related Work
Traditions in ontology versioning [17] and ontology evolution [19] are interested in
finding mappings between different ontology versions, doing ontology refinements and
other changes in the conceptualization [16,22], and in reasoning with multi-version
ontologies[8]. In ontology mapping research, there have been efforts to do mappings
based on probabilistic frameworks [21]. Means for handling inconsistencies between
ontology versions [5] have been developed. Methods for modeling temporal RDF have
been proposed recently [4].
In contrast to these works, our approach is merely about the evolution of ontology
time series that is due to changes in the underlying domain. Hence it should not be
confused with ontology versioning, database evolution, or ontology evolution even if
changes are considered in all of these approaches as well. Each temporal member on-
tology in a time series is a valid, consistent model of the world within the time span it
concerns, and may hence be used correctly in e.g. annotation.
6.3 Future Work
In the future, we would like to investigate whether the methods and tools presented in
this paper could be generalized to other domains, where concepts overcome changes

14
affecting their extensions, properties, or positions in ontological hierarchies and struc-
tures.
Acknowledgements
Our research is a part of the National Semantic Web Ontology Project in Finland18
(FinnONTO, 2003–2007 and 2008–2010) funded by the Finnish Funding Agency for
Technology and Innovation(Tekes) and a consortium of 38 companies and public orga-
nizations.
References
1. D. Brickley and R. V. Guha. RDF Vocabulary Description Language 1.0: RDF Schema W3C
Recommendation 10 February 2004. Recommendation, World Wide Web Consortium, Feb.
2004.
2. J. J. Carroll, I. Dickinson, C. Dollin, D. Reynolds, A. Seaborne, and K. Wilkinson. Jena:
Implementing the semantic web recommendations. Technical Report HPL-2003-146, HP
Labs, December 24, 2003.
3. M. de Berg, M. van Kreveld, M. Overmars, and O. Schwarzkopf. Computational geometry:
algorithms and applications. Springer-Verlag New York, Inc., Secaucus, NJ, USA, second
edition, 2000.
4. C. Gutierrez, C. Hurtado, and A. Vaisman. Temporal RDF. In 2nd. European Semantic Web
Conference (ESWC2005), pages 93–107, Heraklion, Greece, May 29 - June 1 2005.
5. P. Haase, F. van Harmelen, Z. Huang, H. Stuckenschmidt, and Y. Sure. A framework for
handling inconsistency in changing ontologies. In International Semantic Web Conference,
pages 563–577. Springer, Galway, Ireland, November 6-10,2005.
6. M. Hearst, A. Elliott, J. English, R. Sinha, K. Swearingen, and K.-P. Lee. Finding the flow
in web site search. CACM, 45(9):42–49, 2002.
7. M. Hildebrand, J. van Ossenbruggen, and L. Hardman. /facet: A browser for heterogeneous
semantic web repositories. In The Semantic Web - Proceedings of the 5th International
Semantic Web Conference2006, pages 272–285, November 5-9 2006.
8. Z. Huang and H. Stuckenschmidt. Reasoning with multi-version ontologies: A temporal
logic approach. In International Semantic Web Conference, pages 398–412, 2005.
9. E. Hyvönen. Semantic portals for cultural heritage. In Manuscript draft for a book chapter,
Sept 2007.
10. E. Hyvönen, M. Junnila, S. Kettula, E. Mäkelä, S. Saarela, M. Salminen, A. Syreeni, A. Valo,
and K. Viljanen. Finnish Museums on the Semantic Web. User’s perspective on museumfin-
land. In Selected Papers from an International Conference Museums and the Web 2004
(MW2004), Arlington, Virginia, USA, 2004.
11. E. Hyvönen, R. Lindroos, T. Kauppinen, and R. Henriksson. An ontology service for geo-
graphical content. In Poster+Demo Proceedings, ISWC+ASWC-2007, Busan, Korea, 2007.
12. E. Hyvönen, T. Ruotsalo, T. Häggström, M. Salminen, M. Junnila, M. Virkkilä, M. Haaramo,
T. Kauppinen, E. Mäkelä, and K. Viljanen. CultureSampo—Finnish culture on the Semantic
Web: The vision and first results. In Semantic Web at Work — Proceedings of the 12th
Finnish Artificial Intelligence Conference STeP 2006, volume 1, Helsinki, Finland, 2006.
18 http://www.seco.tkk.fi/projects/finnonto/

15
13. T. Kauppinen, C. Deichstetter, and E. Hyvönen. Temp-o-map: Ontology-based search and
visualization of spatio-temporal maps. In Demo track at the European Semantic Web Con-
ference ESWC 2007, Innsbruck, Austria. Springer, June 4-5 2007.
14. T. Kauppinen and E. Hyvönen. Modeling coverage between geospatial resources. In Posters
and Demos of the 2nd European Semantic Web Conference ESWC2005, Heraklion, Greece,
May 29 - June 1 2005. European Semantic Web Conference. to appear.
15. T. Kauppinen and E. Hyvönen. Ontologies: A Handbook of Principles, Concepts and Appli-
cations in Information Systems, chapter Modeling andReasoning about Changes in Ontology
Time Series. Springer-Verlag, 2007.
16. M. Klein. Change Management for Distributed Ontologies. PhD thesis, Vrije Universiteit
Amsterdam, Aug. 2004.
17. M. Klein and D. Fensel. Ontology versioning on the Semantic Web. In Proceedings of the
International Semantic Web Working Symposium (SWWS), pages 75–91, Stanford University,
California, USA, July 30 – Aug. 1, 2001.
18. E. Mäkelä, E. Hyvönen, and T. Sidoroff. View-based user interfaces for information retrieval
on the semantic web. In Proceedings of the ISWC-2005 Workshop End User Semantic Web
Interaction, Nov 2005.
19. N. Noy and M. Klein. Ontology evolution: Not the same as schema evolution. Knowledge
and Information Systems 5, 2003.
20. E. Oren, R. Delbru, and S. Decker. Extending faceted navigation for rdf data. In International
Semantic Web Conference, pages 559–572, November 5-9 2006.
21. R. Pan, Z. Ding, Y. Yu, and Y. Peng. A bayesian network approach to ontology mapping.
In International Semantic Web Conference 2005, pages 563–577. Springer, Galway, Ireland,
November 6-10,2005.
22. L. Stojanovic. Methods and Tools for Ontology Evolution. PhD thesis, University of Karl-
sruhe, Germany, 2004.
23. H. Stuckenschmidt and F. V. Harmelen. Information Sharing on the Semantic Web. Springer-
Verlag, Berlin Heidelberg, New York, 2004.
24. U. Visser. Intelligent information integration for the Semantic Web. Springer-Verlag, Berlin
Heidelberg, New York, 2004.