=Paper=
{{Paper
|id=Vol-443/paper-2
|storemode=property
|title=Implementation and Evaluation of a Semantics-based User Interface for Web Gazetteers
|pdfUrl=https://ceur-ws.org/Vol-443/paper2.pdf
|volume=Vol-443
}}
==Implementation and Evaluation of a Semantics-based User Interface for Web Gazetteers==
https://ceur-ws.org/Vol-443/paper2.pdf
Implementation and Evaluation of a Semantics-based User
Interface for Web Gazetteers
Krzysztof Janowicz Mirco Schwarz Marc Wilkes
Institute for Geoinformatics Institute for Geoinformatics Institute for Geoinformatics
University of Münster, University of Münster, University of Münster,
Germany Germany Germany
janowicz@uni-muenster.de mirco.schwarz@uni-muenster.de marc.wilkes@uni-muenster.de
ABSTRACT techniques developed in conjunction with the social Web
As the Web moves towards structured data, new communi- 2.0, such as search-while-you-type forms (also known as au-
cation paradigms and user interfaces become necessary to tocompletion) and tag clouds, with the reasoning capabilities
support users beyond simple keyword search. In contrast of the semantic Web.
to applications for expert users, these interfaces should hide
the complexity of ontology-based systems but still offer their GAZETTEERS AS APPLICATION AREA
reasoning capabilities. Using Web gazetteers as application Gazetteers are place name directories. Each gazetteer record
area, we discuss how subsumption and similarity reasoning consists at least of a triple (N, F, T ) where N corresponds
can be integrated into modern Web interfaces and explain the to one or more place names, F represents one or more geo-
made design decisions. To evaluate our approach, the results graphic footprints (i.e., locations), and T is the type (class)
from a first human participants test are presented. of the described feature (i.e., the representation of a real
world geographic entity). Hence, gazetteers support at least
Author Keywords two types of queries. First, they map between place names
Information Retrieval, Similarity, Geographic Feature Types and respective footprints: N → F ; and second, they map be-
tween names and geographic feature types: N → T . Note
ACM Classification Keywords that a named geographic place is an abstract individual de-
H.5.2 Information Interfaces and Presentation: User Inter- fined to refer to a physical region in space and categorized
faces—Graphical user interface; H.3.3 Information Storage according to commonly agreed characteristics. The place
and Retrieval: Information Search and Retrieval name is a handle to support communication. Hence, Place
is a social construct of interest for a particular community
INTRODUCTION AND MOTIVATION during a specific time span. A place may change its name,
The major benefit of semantics-based information retrieval location, and type (e.g., from city to ruin) over time [10, 12].
is the possibility to reach beyond simple keyword search,
and hence to support the user in answering complex queries. Gazetteers are key components of all georeferenced infor-
While there are several promising approaches to semantics- mation systems, including GIScience applications in many
based retrieval such as subsumption and similarity reason- diverse fields, Web-based mapping services such as Google
ing, there is surprisingly little work on how to hide the ad- MapsTM , and an increasing number of Web 2.0 applications
ditional complexity (e.g., the underlying ontology) from the such as plazes.com. Gazetteers are crucial for geoparsing
user. In addition, the results have to be presented in a way where references to geographic locations (by their names)
that they are also comprehensible for non-experts. The ex- are recognized in texts and converted to coordinate refer-
tended retrieval capabilities of the semantic Web should not ences [10]. Most gazetteers are either accessed via a Web
be slowed down by more complex user interfaces. page or through an application programming interface (API).
The geographic feature type is selected from a typing scheme,
Based on our previous work [14, 13], we discuss the im- in most cases from a feature type thesaurus. The type of the
plementation and evaluation of a semantics-based user in- described geographic feature is of special importance. For
terface for Web gazetteers. It makes use of subsumption instance, a search for places named Berlin might return more
and similarity reasoning to support the users in navigating than hundred places with a lot of different feature types vary-
through geographic feature types. Our idea is to combine ing from capital (located in Germany), over administrative
area (a state in Germany), to stream (located in Colombia).
One of the major Web gazetteers is the Alexandria Digital
Library (ADL) Gazetteer1 . The ADL Web interface depicted
in figure 1 includes a map to narrow down the search to a par-
ticular spatial extent, a form to enter the place name (or parts
of it), as well as a scrolldown list with about 200 (or 1200
Workshop on Visual Interfaces to the Social and the Semantic Web 1
(VISSW2009), IUI2009, Feb 8 2009, Sanibel Island, Florida, USA. http://www.alexandria.ucsb.edu/gazetteer/
Copyright is held by the author/owner(s).
1
if one includes the non-preferred) geographic feature types. SEMANTICS-BASED INFORMATION RETRIEVAL
To learn about the definitions of these types, the user has to In contrast to mere syntactic search, semantics-based infor-
switch to the ADL Feature Type Thesaurus (FTT) and read mation retrieval takes the underlying conceptualization (or
the (often ambiguous and brief) plain text descriptions. The attribution) into account to improve searching and brows-
thesaurus also contains information about super types and ing through structured data. Three approaches can be dis-
some manually selected related types (with many types re- tinguished: 1) those based on classical subsumption reason-
maining unrelated to others). Some gazetteers try to take up ing, 2) those based on so-called non-standard inference tech-
the participantion idea of the Web 2.0 and allow for user de- niques such as computing the least common subsumer (lcs)
fined places and types. In both cases, the navigation through or most specific concept (mcs) [18, 17], and finally 3) those
and understanding of the types is a major shortcoming [12]. approaches based on computing semantic similarity2 [5, 6,
3, 14]. As depicted in figure 2, subsumption reasoning can
The following example illustrates the navigation between be applied to vertical search, while similarity works best for
feature types in ADL. A user selects lakes from the feature horizontal search and should not be applied to compare sub
type scrolldown list using the ADL gazetteer Web interface. and super types.
Lakes is a narrower term [2] of hydrographic features which
is a top term in the FTT hierarchy. As the user cannot find
a particular geographic feature, she decides to change the
type to search for reservoirs. Unfortunately, reservoirs is
not a narrower term of hydrographic features, and therefore
she looks up the textual definition of lakes in the thesaurus.
Here, reservoirs is specified as related term [2] and she can
click on the link and read the definition of reservoirs. It is
defined as narrower term [2] of hydrographic structure. This
in turn is a narrower term of manmade features which is an-
other top term in the ADL FTT.
Figure 2. Vertical and horizontal retrieval within an ontology.
Formally, the result set for a user’s query using subsumption
reasoning is defined as RS = {C | C v Cs }; where Cs is
the search (or query) concept specified by the user. As each
concept returned to the user is a subsumee of the search con-
cept, it meets the user’s search criteria. This makes selecting
an appropriate search concept the major challenge for end
users. In fact, the search concept is an artificial construct
and not necessarily the searched concept [15]. If it is too
generic, i.e., at the top of the hierarchy, the user will get a
large part of the queried ontology back as (unsorted) result
set. In contrast, if the search concept is too specific, the re-
Figure 1. The Alexandria Digital Library Gazetteer Web interface sult set might be empty.
showing a search for reservoirs in Florida.
In case of similarity-based retrieval, the result set is defined
As discussed by Janowicz and Keßler [12], semantics-based
as RS = {Ct | sim(Cs , Ct ) > t}; where Ct is the compared-
user interfaces in conjunction with geographic feature type
to target concept and t a threshold defined by the user or ap-
ontologies would support the user in browsing through and
plication [14, 15]. In contrast to subsumption reasoning, the
searching in gazetteers. The alternative interface proposed in
search concept is the concept the user is really searching for
this paper aims at hiding the feature type scrolldown list (see
(no matter if it is part of the queried ontology or not). As
figure 1) from the user. Instead, we propose a search-while-
the concepts are compared for overlap (of their definitions
you-type form as well as a combination of horizontal and
or extensions), similarity is more flexible than subsumption
vertical search to improve the navigation between (related)
reasoning, but it is not guaranteed that the results match all of
feature types. These types will be determined based on their
the user’s search criteria. Note that the similarity estimations
formal specifications. Hence, users do not have to look up
between search and target concepts are asymmetric which is
potentially relevant types in the FTT by hand.
important for information retrieval. Usually, the results of a
2
We are focussing on measures which define similarity as degree
of conceptual overlap here. Other theories consider similarity as
(inverse) distance within a multi-dimensional space, or as a set of
transformations [7] from the search to the target concept.
2
similarity query are presented to the user as an ordered list
with descending similarity values.
Summing up, the benefits similarity offers during the re-
trieval phase, i.e., to deliver a flexible degree of (conceptual)
overlap to a searched concept, stands against shortcomings
during the usage of the retrieved information, namely that
the results not necessarily fit the user’s requirements.
To overcome these shortcomings, similarity theories such as
SIM-DL combine subsumption and similarity reasoning. In
a first step, the context of discourse [11] is defined by intro-
ducing a so-called context concept such that Cd = {Ct |Ct v
Cc }. Consequently, in the next step only such concepts are
compared for similarity which are subconcepts of Cc . This
way, the user can specify some minimal characteristics all Figure 3. Combining subsumption and similarity reasoning for in-
target concepts need to share. As depicted in figure 3, Cti formation retrieval. Cs is the search concept, Cti and Ctj are the
compared-to target concepts, while Cx is not a subsumee of the con-
and Ctj are compared for similarity to Cs while Cx is not. text concept Cc , and hence is not compared for similarity to Cs .
Note that for reasons of simplification, the figures 2 and 3
show a single hierarchy, while similarity takes all role-filler
pairs (e.g., the fact that a river has a spring as its origin content or the type of data. Van Kleek and colleagues [24]
∃ hasOrigin.Spring) into account to compute conceptual point to the obstacles users face when diving into the seman-
overlap. tic Web and introduce an agenda on how to improve knowl-
edge creation and access for end users. Van Ossenbruggen et
With regard to semantics-based user interfaces, subsumption al. argue why it is difficult to evaluate semantics-based user
reasoning can be applied to include lakes in the result set if interfaces [25]. Basically, one needs to distinguish between
the user is searching for (the string) waterbody. Mcs and lcs the quality of the underlying data, the involved reasoning
can be used to implement a query-by-example in where the engine, and the user interface as such. In this work, we fo-
result set is computed out of a set of prototypical individuals cus on the user interface. Additional information on the un-
(e.g., waterbodies) selected by the user [26, 15]. Similarity derlying data and ontologies is discussed in [12], while the
measures can be implemented to propose alternative or sup- similarity reasoning is described in [14].
plementary results such as reservoirs if the user is searching
for (the string) lake. However, to propose meaningful alter- The Atom interface [21] realizes a circular layout to support
natives the underlying similarity measure needs to be cog- users in navigating through semantic Web data. One exam-
nitively plausible, i.e., the rankings returned by such a mea- ple for a combination of Web 2.0 interaction and semantic
sure have to correlate with human similarity judgments for Web techniques is given by Heath and Motta [8], who de-
the same set of compared individuals or concepts based on scribe how non-experts can publish and consume RDF data
the same definitions3 . This has been shown for several mea- by means of tags in a reviewing Website. Whereas their
sures such as MDSM [19] and SIM-DL [14, 13]. As these work deals with tagging data and proposing related tags, we
similarity measures compare DL concepts defined within on- present an approach to interact with data classified in a tax-
tologies, the question what is similar to Lake depends on the onomy (of feature types) and focus on similarity and sub-
formal definition (and does not necessarily match the mental sumption instead of relatedness. A promising approach to
conceptualization of a particular human user). integrate semantic autocompletion for geographic locations
is described by Hildebrand et al. [9]. They propose an inter-
SEMANTICS-BASED USER INTERFACES face which allows for searching geographic places by name
While there are specifications on how to display RDF triples where the suggestions are grouped by country or feature
to the user [4, 20] and various semantic Web browsers such type. Instead of focusing on a search by name, in this paper,
as PowerMagpie4 , there is little work on how to integrate the we implement an additional feature type parameter, allowing
reasoning capabilities of the semantic Web within end user for a more restrictive search of places but also demanding for
interfaces for information retrieval. Schraefel and Karger an intuitive feature type selection process. An approach to
[22] argue against the assumption that data on the semantic ease concept selection is presented by Sinkkilä et al. [23].
Web should be displayed to end users as graphs simply be- They present an interface that combines semantic autocom-
cause graphs are used for the computational representation pletion with cross-ontological context navigation. Whereas
(as data model). As in contrast to HTML, XML (and there- Sinkkilä et al. chose to visualize the concept hierarchy as
fore RDF and OWL) separates design from content, the vi- well as related concepts, we suggest to hide the taxonomic
sualization should depend on the task to solve instead of the complexity from the user.
3
This is a weaker requirement than cognitive adequacy which
would imply that the computational similarity theory reproduces
the process of human similarity reasoning.
4
http://powermagpie.open.ac.uk/
3
IMPLEMENTATION and opacity visualize the ranked order of the similarity val-
In order to demonstrate and evaluate how semantics-based ues, where a big font size and high opacity indicate a high
information retrieval can be integrated in Web interfaces for similarity. Each feature type entry in the table is a hyper-
end users, we have implemented a prototypical interface for link. If a user clicks on one of the suggested feature types,
the ADL gazetteer. The interface can access the records it is selected as search parameter and the similar and super
stored in the gazetteer via the ADL API. To implement the types are computed thereupon. The color of the chosen fea-
search-while-you-type form as well as geographic feature ture type changes to orange to indicate that this type is part
type recommendations, the prototype makes use of AJAX of the new query.
(Asynchronous JavaScript and XML) which allows to up-
date certain parts of the Web interface without the need to The features resulting from a specific query are displayed at
reload the whole page. the right side of the interface and on the map. Clicking on
these links or map markers opens a text box which contains
As depicted in figure 4, the gazetteer Web interface allows the information from the ADL gazetteer (e.g., names, foot-
for specifying three search parameters: place name, geo- prints, broader administrative units).
graphic feature type, and location of the searched place. As
argued above, while specifying the name and location of a Figure 4 shows the result of an exemplary search for places
searched place is straightforward, choosing the correct fea- of type Reservoir located in southern Florida (without name
ture type is more difficult, especially in combination with the restriction). Using the interface, the search could be broad-
other parameters. An example is shown in figure 4; note that ened by clicking on the supertype Waterbody, leading to a
most of the places categorized as reservoirs are named lake larger result set (including the results for Reservoir). In case
or pond, hence they would not appear in the result list if the the search for reservoirs did not yield the desired result, the
user would search for geographic features of type Lake. user could also decide to search for similar geographic fea-
ture types such as Lake or Canal.
To point out why a semantics-based interfaces is valuable for
end users, imagine the following example. One of the search
results shown in figure 4 is Lake Manatee. Due to its name,
a user searching for this specific place might be tempted to
deduce that it is of type Lake. Specifying a query as shown
in figure 6 seems obvious, but will not return Lake Manatee,
since it is of type Reservoir. However, facing an empty result
set, the user could decide to either broaden the search by
selecting the supertype Waterbody, or try out a similar type
to Lake. In the latter case, searching for Reservoir, which is
most similar to Lake, yields the desired result.
EVALUATION
In this section we discuss first results and lessons learned
Figure 4. The semantics-based Web interface for the ADL gazetteer
showing the search result for reservoirs in southern Florida.
from a human participants test carried out using the new in-
terface.
When the users begin to enter characters in the feature type Our hypothesis was that an interface combining subsump-
input field a table appears showing suggested types, starting tion and similarity to support the user in selecting appro-
with the characters already entered in the input field. There- priate feature types would be more effective than interfaces
fore, users can immediately see which feature types are sup- solely based on either subsumption or similarity reasoning.
ported by the application without browsing the feature type We did not compare our interface to the original ADL inter-
thesaurus (or ontology) manually. In addition, the table pro- face. This is for the following reasons:
vides information about similar types and super types (which
are computed by the SIM-DL similarity server5 at query • The original ADL interface makes use of the ADL Fea-
time). These recommendations support the user in browsing ture Type Thesaurus while the new interface is based on
the underlying feature type ontology (vertically and horizon- a feature type ontology which contains only hydrographic
tally; see figure 3). feature types so far [12]. Additionally, we do not distin-
guish between preferred and non-preferred concepts and
To display the similarity rankings in the Web interface font have one root node instead of six.
size and opacity are used. This is known from tag clouds, as
used by several blogs or Web 2.0 recommendation portals,
and should therefore be familiar for the user. The font size • The new interface supports search-while-you-type, while
the ADL interface does not. Consequently, the partici-
5
The SIM-DL similarity server as well as the interface are free pants would have to scroll through the whole ADL feature
and open-source software and can be downloaded at http://sim- type list – while only some of these types are relevant for
dl.sourceforge.net/. the test.
4
• The ADL interface allows for multiple comparison oper- given to the participants. They were asked to speak out loud
ators for the place names (has any words, has all words, while filling in the test and each test was recorded on video.
has phrase, equals, matches pattern). To reduce complex- To make sure that the tasks are not biased towards subsump-
ity for end users, the new interface only supports has all tion or similarity reasoning, the questionnaire was designed
words for comparison. in a way that it involves both kinds of reasoning as well as in-
teraction with the map and the place name field. Finally, the
• The ADL interface map does not support map markers as participants could give general comments on whether they
used in the Google MapsTM mashup in the new interface liked the interface and how it could be improved.
to narrow down the search to a particular map extent.
As depicted in figure 4, all interfaces were designed in a way
To test the hypothesis we developed three versions of the that the search parameters form a sentence of the type
semantics-based interface, the combined, a similarity-based
and a subsumption-based version. While the subsumption- You are currently looking for places with . . . in the
based interface shows super and sub types, the combined in- place name, and . . . as place type, and that are located
terface only lists the super types (see figure 5). The underly- within this region: . . . .
ing assumption was that users will select specific types rather
than general ones and can use the super types to broaden
their search. Strictly speaking, this could be another hypoth-
esis for further testing. Showing similar, super, and sub types
may overload the interface and overstrain the user.
Figure 5. The geographic feature type recommendation part of the
interfaces: (1) combined interface, (2) similarity-based interface, (3)
subsumption-based interface.
To verify the hypothesis we investigated how many tasks Figure 6. The interface shows the sentence resulting from the user’s
were successfully solved per interface and how many user search parameters while querying.
interactions (e.g., clicks) were necessary. As the interface
acts as front-end for the ADL gazetteer (whose response During a series of pre-tests it turned out that many partici-
times differ), the time needed to solve the tasks cannot be pants are so accustomed to single keyword search fields (as
taken as criterion. A total of 30 participants performed the used in most common Web search engines7 ) that they put the
test6 , i.e., each interface was tested by 10 participants. To place name as well as the feature type into the place name
ensure that the participants understand how the interfaces field. This was still the case when they were asked to speak
work, a video was shown which stepwise presents how to out the query before submitting it. Therefore, the interfaces
solve an introductory task: were redesigned to display the sentence resulting from the
entered search parameters in the map area while loading (and
Ems-Task the participants were still asked to speak out the query). This
reduced the number of wrong queries.
• Find the river Ems in Germany. Make it the only
result shown in the ’found places’ list. Try using Analyzing the first results from our human participants test,
just the ’place name’ field. Try using both fields, it turns out that from a total of 40 tasks per interface, 72.5%
’place name’ AND ’place type’. were solved completely (i.e., including all sub tasks) using
the combined interface. The participants using the similarity-
• Find canals connected to the Ems. Try using based interface were able to solve 65% of the tasks, while
just the ’place name’ field. Try using both fields, 62.5% of the tasks were solved using the subsumption-only
’place name’ AND ’place type’. (Note that the interface.
number of ’found places’ differs. Do you see
why?) As depicted in figure 7, there is no clear difference in the
total number of user interactions necessary to solve the tasks.
Next, the participants were asked to solve the task on their However, the inter-quartile range differs clearly. It increases
own. Afterwards, four additional tasks (with sub tasks) were from the combined interface over the subsumption-based to
6
With an age ranging from 21 to 30 years, 9 were female and the similarity-based interface.
21 male (most of them students of either geoinformatics or com-
7
puter science). The questionnaire and the introductory video can In contrast to upcoming semantics-based search engines such as
be downloaded at http://sim-dl.sourceforge.net/downloads/. Powerset (http://www.powerset.com/).
5
In the current version of the interface, only ten entries of the
result set and their corresponding map markers are shown at
a time. Although it is possible to go forth (or back) to see
the next (or previous) ten entries, several participants stated
that they almost overlooked the fact that there are more than
ten result entries. As a consequence, they proposed to show
all markers on the map, where those currently not listed in
the sidebar could be greyed out. Some participants also re-
quested that features of similar types should be directly dis-
played on the map using colored markers to visualize their
similarity instead of the suggestion table.
Finally, a few participants mentioned that upon sending a
query to the server and reading the query summary, they re-
alized a mistake in one of the search parameters, and hence
missed a button to cancel the request. As this was not possi-
ble, the wrong query was counted as user interaction during
the test.
Figure 7. The box plot shows the median as well as the 0.25 and 0.75
quartiles for the number of total user interactions (clicking, typing in CONCLUSIONS AND FURTHER WORK
words, following suggestions, interacting with the map) per interface In this paper we discussed results from testing the integra-
type. Comb is the combined interface using subsumption and similarity,
while Sim only shows similar types as suggestions and Sub only sub and
tion of subsumption and similarity reasoning into end user
super types. interfaces to support horizontal and vertical retrieval. While
a combination of both seems to be the most promising ap-
proach, it is not surprising that there is a gap between being
While these first results support our hypothesis, the test also able to implement more intelligent user interfaces and mak-
points so some aspects which let us rethink the design of ing them intuitively useable [25]. One solution would be
the Web interface as well as the number of user interac- to hide even more of the complexity (e.g., the suggestions
tions as quality measure. Comparing manually entered ver- table) from the user by directly displaying features of very
sus clicked feature types, intuitively one would expect that similar types. However, this may be interpreted as pater-
the more types are suggested, the more will be clicked in- nalism and reduce the user’s trust in such interfaces. For
stead of being manually entered. Hence, the number of man- instance, while reservoirs and lakes might be similar enough
ually entered feature types should be lower in the combined for most tasks, some application areas such as water man-
interface compared to the other ones. This turns out to be agement may require a clear distinction between both. The
a wrong assumption. In fact, the ratio between clicked and extended retrieval capabilities of the semantic Web should
manually entered types was about one to one in case of the be used to offer support, while the final decision should still
combined and the similarity interface, while being higher be up to the user. This leads to the question of how to in-
for the subsumption-based interface. To understand these corporate contextual information in the user interfaces and
results, we reviewed the videos taken from the participants to which degree context influences the resulting suggestions
during the test. It turns out that several participants had en- [1, 16, 11]. From the viewpoint of the social Web, one could
tered the type manually after seeing it in the suggestion table. also think of allowing human users to add similar types by
This points towards two difficulties. First, we have to inves- hand to improve the suggestions.
tigate whether this was caused by the interface design, i.e.,
by the positioning of the forms and the table (for instance
using eye tracking). Second, this makes it more difficult to
understand to which degree the suggested types were helpful
to the user in selecting the next type to be displayed.
After completing the tasks, the participants were asked to an-
swer additional questions about the design and functionality
of the interface. While the participants liked the interfaces
in general and rated the type suggestion table to be useful
(giving both the second best median rank on a 5-level Likert Figure 8. Conceptual design for the feature type suggestions allowing
scale), it turns out that some of the participants still had prob- for vertical and horizontal navigation.
lems dealing with two input fields. A given suggestion for
improvement was to have just one input field and automat- Based on our experiences and the test results, we designed
ically identify keywords representing feature types. Some a new navigation component which will be integrated in the
participants recommended to cache previous queries, allow- next version of the Web interface (figure 8). This component
ing for an easy comparison of several result sets using the will replace the feature type suggestion table in the upper
browser’s navigation (back and forward) buttons. right part of the interface. It will appear as soon as a feature
6
type is selected or typed in by the user. The suggestion ta- Workshop on Ontology Matching (OM2007) at
ble itself will appear just below the feature type input field ISWC/ASWC2007, Busan, South Korea, 2007.
as usually implemented in autosuggestion forms. The new
component visualizes explicitly the difference between hor- 4. C. Bizer, R. Lee, and E. Pietriga. Fresnel - Display
izontal and vertical navigation. A vertical slider can be used Vocabulary for RDF. Technical report, The World Wide
to specify more general or more specific feature types than Web Consortium (W3C), June 2005.
the searched type. The result set only contains places that
5. A. Borgida, T. Walsh, and H. Hirsh. Towards
are of a type under the slider. The horizontal slider allows
Measuring Similarity in Description Logics. In
for including similar types in the result set. This integrates
Proceedings of the 2005 International Workshop on
the idea of a similarity threshold which was discussed in the
Description Logics (DL2005), volume 147 of CEUR
semantics-based information retrieval section. Here, only
Workshop Proceedings. CEUR, Edinburgh, Scotland,
those places are included in the result set that are to the left of
UK, 2005.
the slider. As in the presented user interface, less opacity in-
dicates decreasing similarity. The use of font size, however, 6. C. d’Amato, N. Fanizzi, and F. Esposito. A Semantic
differs. Whereas previously the font size also represented Similarity Measure for expressive Description Logics.
the similarity value, within the conceptual navigation com- In CILC 2005, Convegno Italiano di Logica
ponent the font size indicates the number of features found Computazionale, Rome, Italy, 2005.
within the map extent for that specific type; where a larger
font size points to a higher number of found places. 7. R. L. Goldstone and J. Son. Similarity. In K. Holyoak
and R. Morrison, editors, Cambridge Handbook of
In figure 8, the immediate super type of Reservoir is Water- Thinking and Reasoning. Cambridge University Press,
body, although another immediate super type of Reservoir 2005.
specified in the underlying ontology is Manmade. Allow-
ing for generalizing also towards Manmade now yields two 8. T. Heath, E. Motta, T. Heath, and E. Motta. Ease of
problems. First, the complexity of the navigation component Interaction Plus Ease of Integration: Combining Web 2.
would increase, ending up in a graph-like representation of 0 and the Semantic Web in a Reviewing Site. Web
the ontology (which is the opposite of our design goal). Sec- Semantics: Science, Services and Agents on the World
ond, generalizing towards multiple super types will result in Wide Web, 6(1):76–83, 2008. Semantic Web and Web
a large and diverse result set. That is why we propose to only 2.0.
generalize along one inheritance path. Deciding which path
of inheritance is shown in the navigation component depends 9. M. Hildebrand, J. van Ossenbruggen, A. Amin,
on the context concept Cc whose integration in the interface L. Aroyo, J. Wielemaker, and L. Hardman. The Design
is an open issue. Space of a Configurable Autocompletion Component.
Technical report, CWI technical report INS-E0708,
November 2007.
ACKNOWLEDGEMENTS
The comments from three anonymous reviewers provided 10. L. L. Hill. Georeferencing: The Geographic
useful suggestions to improve the content and clarity of the Associations of Information (Digital Libraries and
paper. This work is funded by the Semantic Similarity Mea- Electronic Publishing). The MIT Press, 2006.
surement for Role-Governed Geospatial Categories (SimCat)
project granted by the German Research Foundation (DFG 11. K. Janowicz. Kinds of Contexts and their Impact on
Ra1062/2-1), as well as the International Research Training Semantic Similarity Measurement. In 5th IEEE
Group on Semantic Integration of Geospatial Information Workshop on Context Modeling and Reasoning
(DFG GRK 1498). (CoMoRea’08) at the 6th IEEE International
Conference on Pervasive Computing and
Communication (PerCom’08), pages 441–446, Hong
REFERENCES Kong, March 2008. IEEE Computer Society.
1. R. Albertoni and M. D. Martino. Semantic Similarity of
Ontology Instances Tailored on the Application 12. K. Janowicz and C. Keßler. The Role of Ontology in
Context. In A. K. Dey, B. Kokinov, D. Leake, and Improving Gazetteer Interaction. International Journal
R. Turner, editors, ODBASE - OTM Conferences, of Geographical Information Science,
volume 4275 of Lecture Notes in Computer Science, 10(22):1129–1157, 2008.
pages 1020–1038. Springer, 2006.
13. K. Janowicz, C. Keßler, I. Panov, M. Wilkes,
2. ANSI/NISO. ANSI/NISO Z39.19 - 2005 Guidelines for M. Espeter, and M. Schwarz. A Study on the Cognitive
the Construction, Format, and Management of Plausibility of SIM-DL Similarity Rankings for
Monolingual Controlled Vocabularies. NISO Press, Geographic Feature Types. In L. Bernard,
2005. A. Friis-Christensen, and H. Pundt, editors, 11th
AGILE International Conference on Geographic
3. R. Araújo and H. S. Pinto. Towards Semantics-Based Information Science (AGILE 2008), Lecture Notes in
Ontology Similarity. In P. Shvaiko, J. Euzenat, Geoinformation and Cartography, pages 115–133,
F. Giunchiglia, and B. He, editors, Proceedings of the Girona, Spain, 5-8. May 2008. Springer.
7
14. K. Janowicz, C. Keßler, M. Schwarz, M. Wilkes, 24. M. Van Kleek, M. Bernstein, P. André, M. Perttunen,
I. Panov, M. Espeter, and B. Baeumer. Algorithm, D. Karger, and M. Schraefel. Simplifying Knowledge
Implementation and Application of the SIM-DL Creation and Access for End-Users on the SW. In CHI
Similarity Server. In Second International Conference 2008 SWUI Workshop, April 2008.
on GeoSpatial Semantics (GeoS 2007), number 4853 in
Lecture Notes in Computer Science, pages 128–145. 25. J. van Ossenbruggen, A. Amin, and M. Hildebrand.
Springer, 2007. Why Evaluating Semantic Web Applications is
Difficult. In Semantic Web User Interaction at CHI
15. K. Janowicz, M. Wilkes, and M. Lutz. Similarity-based 2008: Exploring HCI Challenges, 2008.
Information Retrieval and its Role within Spatial Data
26. M. Wilkes and K. Janowicz. A Graph-based Alignment
Infrastructures. In 5th International Conference on
Approach to Similarity between Climbing Routes. In
Geographic Information Science (GIScience 2008),
First International Workshop on Information Semantics
Park City, Utah, USA, September 2008; forthcoming.
and its Implications for Geographic Analysis (ISGA
16. C. Keßler. Similarity Measurement in Context. In B. M. ’08) held in conjunction with GIScience 2008, the 5th
Kokinov, D. C. Richardson, T. Roth-Berghofer, and International Conference on Geographic Information
L. Vieu, editors, 6th International and Interdisciplinary Science, Park City, Utah, USA, September 23 2008.
Conference, CONTEXT 2007, volume 4635, pages
277–290. Springer, Copenhagen, Denmark, August
2007.
17. R. Küsters. Non-Standard Inferences in Description
Logics, volume 2100 of Lecture Notes in Artificial
Intelligence. Springer, 2001.
18. R. Möller, V. Haarslev, and B. Neumann.
Semantics-Based Information Retrieval. In Proc.
IT&KNOWS-98: International Conference on
Information Technology and Knowledge Systems, 31.
August- 4. September, Vienna, Budapest, pages 49–56,
1998.
19. A. Rodrı́guez and M. Egenhofer. Comparing Geospatial
Entity Classes: An Asymmetric and Context-dependent
Similarity Measure. International Journal of
Geographical Information Science, 18(3):229–256,
2004.
20. L. Rutledge, J. van Ossenbruggen, and L. Hardman.
Making RDF Presentable: Integrated Global and Local
Semantic Web Browsing. In WWW ’05: Proceedings of
the 14th international conference on World Wide Web,
pages 199–206, New York, NY, USA, 2005. ACM.
21. K. Samp, C. Basca, B. McDaniel, and S. Decker. Atom
Interface - A Novel Interface for Exploring and
Browsing Semantic Space. In Semantic Web User
Interaction at CHI 2008: Exploring HCI Challenges,
2008.
22. M. Schraefel and D. Karger. The Pathetic Fallacy of
RDF. In International Workshop on the Semantic Web
and User Interaction (SWUI) 2006, Athens, Georgia,
USA, November 2006.
23. R. Sinkkilä, E. Mäkelä, E. Hyvönen, and T. Kauppinen.
Combining Context Navigation with Semantic
Autocompletion to Solve Problems in Concept
Selection. In K. Belhajjame, M. d’Aquin, P. Haase, and
P. Missier, editors, SeMMA, volume 346 of CEUR
Workshop Proceedings, pages 61–68. CEUR-WS.org,
2008.
8