=Paper=
{{Paper
|id=None
|storemode=property
|title=Aligning Unions of Concepts in Ontologies of Geospatial Linked Data
|pdfUrl=https://ceur-ws.org/Vol-798/paper4.pdf
|volume=Vol-798
}}
==Aligning Unions of Concepts in Ontologies of Geospatial Linked Data==
https://ceur-ws.org/Vol-798/paper4.pdf
Aligning Unions of Concepts in Ontologies of
Geospatial Linked Data
Rahul Parundekar, José Luis Ambite, and Craig A. Knoblock
Information Sciences Institute and Department of Computer Science
University of Southern California
4676 Admiralty Way, Suite 1001, Marina del Rey, CA 90292
{parundek,ambite,knoblock}@usc.edu
Abstract. It is evident from the recent growth in Geospatial Linked
Data that even though the number of instances being generated and
linked has increased drastically, the ontologies behind these sources re-
main disconnected. Though we can agree that the instances being linked
are equivalent, the alignments that are extrapolated from these links
between the concepts may or may not agree with our intuitions. It is
important to investigate how the concepts in the sources are actually
aligned. Our previous work was successful in finding alignments, such as
equivalence and subset relations, between concepts of two sources, us-
ing the instances that are linked as equal. Such alignments need not be
trivial, however, as a concept in the ontology might not have an exact
equivalent class in the other source. In this paper we propose a method
that uses the subset and equivalence relations between restriction classes
found by our previous work to find new alignments, where one (larger)
concept of a source is aligned to the union of multiple (smaller) concepts
from another source. We also show that we can use these alignments to
find inconsistencies and use them to identify the instances that may be
erroneously aligned.
1 Introduction
The Web of Linked Data has seen huge growth in the past few years. As of
September 2010, the size of the Linked Open Data Cloud was about 28.5 billion
triples with around 20.6% of the triples belonging to the geospatial domain.1 As
of June 2009, the cloud had recorded an overall growth of about 300% with 91%
growth in the geospatial domain.2 Out of the 16 geospatial data sources covered
in the September 2010 count, there are around 16.5 million outgoing links to
other sources. The sources of Geospatial Linked Data are most popularly con-
nected using the owl:sameAs property, linking instances that are the same. As
more alignments are generated in the Web of Linked Data at the instance level,
a pattern of inter-linked data arises where the ontologies behind the sources
1
http://www4.wiwiss.fu-berlin.de/lodcloud/state/
2
http://events.linkeddata.org/ldow2011/slides/ldow2011-slides-intro.pdf
2 Rahul Parundekar, José Luis Ambite, and Craig A. Knoblock
remain un-linked. As described in our previous papers on Linking and Build-
ing Ontologies of Linked Data [7] and Aligning Ontologies of Geospatial Linked
Data [6], an extensional technique can be used to generate alignments between
the ontologies behind these sources. In these papers, we introduce a concept of
restriction classes, which is similar to that of single value constraints on property
restrictions of the Web Ontology Language (OWL) to increase the expressivity
of sources with a rudimentary ontology. By looking at the set containment rela-
tionships of the instance sets of these restriction classes, we find equivalent and
subset alignments between the two sources. Though the equivalent alignments
found are precise in finding similar concepts between the two sources, the subset
relations found, though informative, are too numerous to be effectively used.
Reviewing these subset relations we discovered that there are potential equiv-
alent alignments not found by our previous work, linking a larger concept to a
union or aggregation of one or more of its subsets. Using this as motivation,
the work described in this paper builds on the ontology alignment method of
[7]. Picking up where we left off, the approach described in this paper uses the
subset relations as hints to create a union of smaller restriction classes, by virtue
of a common property and restriction classes with only a single property-value
pair, which guides the aggregation and then performs set containment operations
with the larger restriction class from the other source. Using this method, we
explore three Geospatial Linked Data sources - GeoNames, DBpedia, & Linked-
GeoData and try to find new alignments between GeoNames & DBpedia and
LinkedGeoData & DBpedia, where a larger subsuming restriction class from one
source can be explained by an aggregation of smaller restriction classes from the
other source.
The scope of this paper is in the domain of Geospatial Linked Data, where
we find alignments between three sources: GeoNames, DBpedia and LinkedGeo-
Data. We first find equivalences and subset relations as described in our previous
work, and then use these to find the new union alignments. The nature of each of
the three sources investigated is briefly mentioned here and they are described in
more detail in [7]. GeoNames is a geographic source with a flat-file like ontology
where all instances belong to a single concept of Feature and have associated
Feature Class & Feature Code property to identify the instances as mountains,
lakes, etc. Although DBpedia is a Linked Data source that covers domains other
than the geospatial domain, there are a large number of instances from GeoN-
ames linked to those in DBpedia using the owl:sameAs property. We also try
to find alignments between the ontologies behind LinkedGeoData and DBpedia.
RDF data in LinkedGeoData is derived from the Open Street Map initiative and
has links to DBpedia.3
This paper is organized as follows. We first describe briefly our alignment
algorithm from [7] along with the limitations of the results that were generated.
We then explain our approach to finding alignments between a larger concept
from one source and the union set of multiple smaller concepts from the other
source. This is followed by identifying the outliers of these alignments that high-
3
http://linkedgeodata.org/Datasets
Aligning Unions of Concepts in Ontologies of Geospatial Linked Data 3
light the inconsistencies and the instances that are erroneously linked. We then
describe the experimental results that contain the new alignments discovered in
these data sources, along with their outliers. Finally, we describe other related
work and conclude with our observations and future work.
2 Aligning geospatial ontologies on the Web of Linked
Data
The work described in this paper follows our previous work on aligning ontologies
of Linked Open Data, which uses an extensional approach to find alignments
between restriction classes in two different sources. Though the results generated
by our previous algorithm found equivalent alignments between the two sources,
a large number of subset alignments were also found. A pattern was observed
in these results, where a group of concepts from one source were subsets of
the same larger concept from the other source. In many cases these smaller
concepts taken together were able to completely explain the larger source. We
used this insight as motivation for consuming the subset relations, which were
too numerous to be useful by themselves, to find alignments between the larger
concept and the union of the group of concepts. Our approach uses this group of
smaller concepts and introduces a disjunction operator on these subsets to try
to define the common subsuming concept.
2.1 Our previous work on linking and building ontologies of Linked
Data
Ontologies of Linked Data sources can be quite rudimentary. For example, GeoN-
ames only has a single concept (Feature) to which all of its instances belong. On
the other hand, in DBpedia, we find a rich ontology with a hierarchy of concepts
and well-defined properties. In the traditional sense of ontology alignment, we
would have found at most a single alignment between Feature on the GeoNames
side and a similar broad concept from DBpedia. In order to get a richer set of
alignments, we introduced the concept of a restriction class. A restriction class
is a concept that is derived extensionally and defined by the set of instances ob-
tained by restricting a single property to a single value (called a property-value
pair and represented by (pi = vi )) in a source. For example, a restriction class for
schools can be constructed in GeoNames by forming a set of instances that have
their geonames:featureCode restricted to ‘S.SCH ’. This restriction class is repre-
sented as geonames:featureCode=S.SCH. The scope of the definition of a restric-
tion class includes the conjunction operator, which produces a more specialized
set of instances, constructed using two or more restriction classes. Thus, a restric-
tion class {geonames:featureCode=S.SCH & geonames:countryCode=US}, built
from the restriction classes geonames:featureCode=S.SCH and geonames:countryCode-
=US, can be defined by the intersection of the two sets and forms a concept
extensionally described by the set of schools in the US in GeoNames.
Our algorithm aligns restriction classes from two sources, using an exten-
sional technique, as follows. A pre-processing step first performs an inner-join
4 Rahul Parundekar, José Luis Ambite, and Craig A. Knoblock
on the two sources to be aligned based on an instance equivalence property
like owl:sameAs. As inverse functional properties can only result in restriction
classes with a single instance belonging to it, the pre-processing step elimi-
nates them. The crux of the algorithm uses a top-down tree exploration of the
space of alignment hypotheses. At the topmost level, a seed hypothesis is gen-
erated by aligning a restriction class with one property-value pair from the first
source with another restriction class with one property-value pair from the sec-
ond source. At each level in the search space, a new restriction class is formed
from one restriction class of one of the sources by adding another property-value
pair constraint on that restriction class. A new alignment hypothesis is thus
constructed from the new restriction class and the restriction class from the
other source. Each alignment hypothesis is tested for set containment relations
between the intersection set of the restriction classes from both sources. This
is done with the help of two scoring functions - P & R. If r1 and r2 are the
two restriction classes in the alignment hypothesis, we first define Img(r1 ) as
the set of instances in the second source that instances of r1 are linked to. We
then define P as |Img(r 1 )∩r2 |
|r2 | , and R as |Img(r 1 )∩r2 |
|Img(r1 )| . We mark the relation of the
alignment hypothesis as either i) equivalent (P = 1, R = 1), ii) subset, with the
restriction class from the first source as extensionally subsuming the restriction
class from second source (R = 1), iii) subset, with restriction class from second
source extensionally subsuming the restriction class from first source (P = 1) or
iv) no relation between the two restriction classes. To compensate for missing
and misaligned instances, we relax our subset scores by defining P 0 and R0 that
reduce the required fraction of support to be greater than 0.9 instead of equal
to 1. For an optimal exploration of the search tree, we employ certain pruning
mechanisms that include i) using ordered exploration to avoid exploring a node
twice, ii) pruning a node if the intersection set of the restriction classes of the
hypothesis has size less than a minimum support size (we used 10 in our ex-
periements), iii) pruning a node if the added restriction class does not change
the set of instances, etc. After the brute-force exploration of the search space of
alignment hypotheses, we use a post-processing step on the results generated,
which removes redundant assertions by virtue of set containment of instances
of two hypotheses where one is the immediate parent of the other in the search
tree.
At the end of the above three steps of processing, the algorithm was able
to find equivalent relations between restriction classes from two sources as well
as subset relations in either direction. As this algorithm was not specific to
any particular domain, we explored candidate sources for alignments in three
domains: Geospatial, Genetics and Zoology. In these three domains, our algo-
rithm found alignments of 5 pairs of sources. For example, we were able to find
alignments between GeoNames and DBpedia in the Geospatial domain. One such
alignment was the equivalent relation between {geonames:countryCode=ES} and
{dbpedia:country=Spain} (i.e. correctly aligning the concepts for the country
Spain). We also found subset relations like {geonames:featureCode=S.SCH} sub-
set of rdf:type=dbpedia:EducationalInstitution. More such results are described
in [7].
Aligning Unions of Concepts in Ontologies of Geospatial Linked Data 5
Limitations The approach above produced a large number of equivalent align-
ments that gave an exact mapping between the two restriction classes from
the two sources. It also, however, produced a large number of subset rela-
tions that were not as useful. This was mainly because the subset relations,
by themselves, did not contribute to a useful equivalence alignment between
two classes. In all, in the GeoNames and DBpedia alignment, there were 1647
subset relations found. Though it is understandable that in many cases there
might never exist an exact equivalence between two restriction classes, be-
cause they were auto-generated using property-value pairs, we decided to look
for additional useful alignments, if any, that these subset relations might be
able to provide us. For example, in the GeoNames and DBpedia alignment, we
found that {geonames:featureCode=S.SCH}, {geonames:featureCode=S.SCHC}
and {geonames:featureCode=S.UNIV} (i.e. Schools, Colleges and Universities
from GeoNames) are all subsets of {rdf:type=dbpedia:EducationalInstitution}.
Taken individually, though each of these alignments are correct and insight-
ful, they are not particularly useful in understanding the relationships between
GeoNames and DBpedia. Taken together, however, we found that the union of
these three restriction classes completely define rdf:type=dbpedia:EducationalInstitution.
The limitation of our approach was in the expressivity of our restriction classes.
Though it included restriction classes containing single property-value pairs and
the conjunction operator on those restriction classes, it did not include a dis-
junction operator and hence was unable to make use of the subset relations.
2.2 Identifying spatial concept coverings
As explained above, we were able to identify a pattern where a group of restric-
tion classes from one source were aligned as subsets of a common concept from
the other source. By using these alignments as hints, we were able to construct
the union of the smaller restriction classes and detect if the union was able to
define the larger class entirely. The following section describes this method in
detail. In those cases where we are not able to define the larger class entirely,
our approach is also able to find and explain the missing instances (outliers).
Mapping a restriction class from one source with a union of smaller
restriction classes from the other source Since the problem of finding
alignments with conjunctions and disjunctions of property-value pairs of restric-
tion classes is combinatorial in nature, we focus only on subset relations where
both restriction classes have a single property-value pair and where one is a sub-
set of the other. This helps us find the simplest definitions of concepts and also
makes the problem tractable. Alignments generated by our previous work that
satisfy the single property-value pair constraint are first grouped according to
the subsuming restriction classes. We then identify a strategy for selecting the
smaller restriction classes from within such a group to form the union that best
describes the larger restriction class. Since restriction classes are constructed
by forming a set of instances that have one of the properties restricted to a
single value, aggregating restriction classes from the group according to their
6 Rahul Parundekar, José Luis Ambite, and Craig A. Knoblock
properties builds a more intuitive definition of the union. We can now define
the disjunction operator that constructs the union concept from the smaller
restriction classes in these sub-groups. The disjunction operator is defined for
restriction classes, such that i) the concept formed by the disjunction of the
restriction classes represents the union of their set of instances, ii) each of the
restriction classes that are aggregated contain only a single property-value pair
and iii) the property is the same for all those property-value pairs. We then try
to find the alignment between the larger common restriction class and a set of
restriction classes from the other source that are aggregated by the disjunction
operator by using an extensional approach similar to our previous paper. We
call such an alignment as union alignment.
We first build candidates for aggregation using the results from our previ-
ous algorithm as hints. We group alignments by the larger common restriction
class. Grouping the subset relations is trivial. Equivalence relationships are sub-
sets in both directions and thus are easily integrated into the groups. For each
alignment, {p1 =v1 } is the r1 part and {p2 =v2 } forms the r2 part (each with a
single property-value pair ) as explained in the previous section. Sub-groups are
formed by aggregating according to the property of the property-value pairs of
the smaller restriction classes. Such a sub-group is identified by {Property of
the larger restriction class(p1 ), Value of the larger restriction class(v1 ), property
of the smaller restriction classes(p2 )}. Values of the different smaller restriction
classes can be denoted by a list List(v2 s). The disjunction of the smaller re-
striction classescreates a set of instances that extensionally identifies the union
concept. We can now either confirm or refute the hypothesis that the larger
restriction class is equivalent to the union concept. We can do this by using
a scoring mechanism similar to the use of P & R in our previous paper. Us-
ing the same terminology, UA is defined as the set of disjunctive instances (i.e.
Union(Img(r1 ) ∩ r2 ))), UL is defined as the set of instances of the larger class
taken by itself (i.e. Img(r1 )) and US is defined as the set of instances that is
the union of individual smaller restriction classes(i.e. Union(r2 )). The scoring
mechanism defines PU as U UA 0 0
US and RU as UL . PU & RU are defined as fractions
A
0 0
with relaxed scoring assumptions similar to P & R from our previous paper.
For example, our previous algorithm finds that {geonames:featureCode =
S.SCH}, {geonames:featureCode = S.SCHC}, {geonames:featureCode = S.UNIV}
are subsets of {rdf:type=dbpedia:EducationalInstitution}. In this case, the sub-
group can be identified as {rdf:type, dbpedia:EducationalInstitution, geonames:featureCode}
and list as (S.SCH, S.SCHC, S.UNIV). As can be seen in the Venn diagram of
Figure 1, UL is the restriction classImg({rdf:type = dbpedia:EducationalInstitution}),
US is {geonames:featureCode = S.SCH} ∪ {geonames:featureCode = S.SCHC}
∪{geonames:featureCode = S.UNIV} and UA is:
{Img({rdf:type = dbpedia:EducationalInstitution}) ∩ {geonames:featureCode
= S.SCH}} ∪ {Img({rdf:type = dbpedia:EducationalInstitution}) ∩ {geonames:featureCode
= S.SCHC}} ∪ {Img({rdf:type = dbpedia:EducationalInstitution}) ∩ {geonames:featureCode
= S.UNIV}}
Aligning Unions of Concepts in Ontologies of Geospatial Linked Data 7
Ideally, for an exact equivalence alignment, PU0 & RU 0
should both be 1.0,
if the larger restriction class covers the union of the smaller restriction classes
completely and vice-versa. However, similar to the relaxed score assumption
from our previous paper to accommodate errors in the dataset, we consider it a
complete coverage when the score is greater than a relaxed score of 0.9. (i.e. the
union alignment is considered to be equivalent if PU0 > 0.9 & RU 0
> 0.9). Due to
the minimum support score constraint for subsets from our previous paper, we
are assured that U 0 4
US i.e. PU is always going to be greater than 0.9. Thus, we can
A
0
say that a union alignment is equivalent if RU > 0.9. With the educational insti-
0
tutions example, RU for the alignment of dbpedia:EducationalInstitution to the
union of S.SCH, S.SCHC & S.UNIV is 0.98. We can thus confirm the hypoth-
esis and consider this union alignment equivalent. The scores for other union
alignments found are described in the results section.
Key:
Img(r1) : Educational Institutions from Dbpedia
Union(r2): Schools, Colleges and Universities from
Geonames.
S.SCH S.SCHC
Schools from Geonames.
Colleges from Geonames.
S.UNIV
Universities from Geonames.
Img(r1)
Outliers.
EducationalInstitution Union(r2)
Fig. 1. Spatial covering of Educational Institutions from DBpedia
Using mappings to identify outliers As mentioned above, the score for
the alignment of {rdf:type = dbpedia:EducationalInstitution} to the union of
{S.SCH, S.SCHC & S.UNIV} is approximately 0.98. For {rdf:type = dbpe-
dia:EducationalInstitution}, 396 instances out of the 403 Educational Institu-
tions were accounted for as having their geonames:featureCode as one of S.SCH,
S.SCHC or S.UNIV to give this score. An interesting question to pose then
is, how are the remaining 2% of the dbpedia:EducationalInstitutions (i.e. 7 in-
stances) classified in GeoNames?
While calculating the disjuncted restriction classes, we also keep track of
other instances with the same {p1 , v1 , p2 } but not previously considered as sub-
sets. These had been pruned in the exploration stage as they either had a size
of less than the minimum support size constraint of ten instances or had P 0
less than 0.9. For the first type of restriction classes, those with low support
size but yet having P 0 greater than 0.9 are now re-classified as subsets. The
4
It should also be noted that each of the smaller subsets also satisfy the minimum
support size of 10 instances.
8 Rahul Parundekar, José Luis Ambite, and Craig A. Knoblock
re-classification of the relation as a subset can now be justified due to increased
evidence in suggesting subsumption as other values for the same property are
also aligned as subsets of the larger restriction class from the first source.
The second type of restriction classes that had P 0 less than 0.9 along with the
ones that were not re-classified above (i.e. with less than 10 instances and P 0 less
than 0.9) form the outliers. For example, as mentioned before, schools, colleges
and universities from GeoNames make up 396 out of 404 Educational Institutions
from DBpedia. From the other eight instances, 7 have their feature codes as ei-
ther S.BLDG (3 buildings), S.EST (1 establishment), S.HSP (1 hospital), S.LIBR
(1 library) or S.MUS (1 museum). The eighth instance does not have a geon-
ames:featureCode property asserted. The P 0 score of these restriction classes is
less than 0.9. One of the instances classified as dbpedia:EducationalInstitution in
DBpedia is linked to an instance in GeoNames that has geonames:featureCode as
‘S.HSP’. 5 There are 31 instances in {geonames:featureCode=S.HSP}, however,
and because this restriction class does not meet the relaxed subset score thresh-
old, it cannot be considered in the union of restriction classes. Another example
of outliers was found in the {dbpedia:country = Spain ≡ geonames:countryCode
= ES} alignment. This equality was found using the relaxed subset assumption,
where 3917 of the 3918 instances of dbpedia:country=Spain were accounted for as
having geonames:countryCode=ES, resulting in a subset score of 0.9997. The one
instance not having country code ES was actually classified as having country
code IT (Italy). This single instance needs to be inspected further and it needs
to be determined if the owl:sameAs link is correct. It is evident from the above
examples that the outliers help in understanding the nature of the sources more
explicitly, showing why the alignments failed to completely describe the larger
restriction class. These, along with a few other examples, are described in detail
in the next section.
3 Experimental Results
From the approach described in Section 2.2, we were able to get a total of 752
union alignments for the GeoNames-DBpedia alignment and 5843 for the Linked-
GeoData-DBpedia alignment. From the 752 in GeoNames-DBpedia, 318 are such
that the larger restriction class is from DBpedia, while the other 434 have the
larger restriction class from GeoNames. Similarly, 3097 from the 5843 union
alignments in LinkedGeoData-DBpedia have the larger restriction class from
DBpedia, while the other 2746 have the larger restriction class from GeoNames.
Tables 1, 2, 3, & 4 list a few interesting examples of these union alignments
between GeoNames-DBpedia and LinkedGeoData-DBpedia (in either direction),
which we describe here. The tables are organized as follows. Column 2 describes
the sub-group, i.e. (p1 ,v1 ,p2 ). Column 3 contains the list of the value part of the
property-value pairs in the restriction classes of the smaller sets (i.e. List(v2 )).
The score of the union is noted in column 4 (RU 0
= |U A|
|UL | ) followed by |UA | and
5
Intuitively, it would make sense to the reader that this instance might perhaps be a
hospital of a medical school.
Aligning Unions of Concepts in Ontologies of Geospatial Linked Data 9
|UL | in columns 5 and 6. Column 7 describes the outliers, i.e. values of v2 that
form restriction classes that aren’t direct subsets of the larger restriction class.
Each of these values also has a fraction with the number of instances that do
belong to the larger restriction class of the total number of instances of the
restriction class (or |Img(r
|r2 |
1 )|
). It can be seen that the fraction is less than our
relaxed subset score. If the value of this fraction was greater than the relaxed
subset score (i.e. 0.9), the set would have been included in column 3 instead.
The last column mentions how many of the total UL instances we were able to
explain using UA and the outliers. For example, the union alignment of #1, is
the Educational Institution example described before. It shows how educational
institutions from DBpedia can be explained by schools, colleges and universities
0
in GeoNames. Column 4, 5 and 6 explain the alignment score RU (0.98), the
size UA (396) and the size of UL (404). The seven of the eight outliers found
(S.BLDG, S.EST, S.LIBR, S.MUS, S.HSP) are mentioned along with their P 0
fractions in column 7.
We also found some other interesting alignments. #2 shows the details of the
Spain example mentioned briefly in Section 2.2. #3 shows a union alignment
that aligns smaller sets or parts from GeoNames to a complete set. The region
of Basse-Normandie in France is made up of three departments. The restric-
tion classes of these three regions are constrained by the geonames:parentADM2
property. #4 shows that Airports and Airbases make up 99% of the airports in
DBpedia. From its outliers, one might argue that Airfields (S.AIRF) should also
be included, but it was not as its P 0 score was lower than the threshold. Outliers
also show that there is a Hill in geonames that has been classified as an airport.
Even though this instance may be an airport in the hills, ontologically it doesn’t
make sense that a hill can be an airport. A similar case is observed in #8 where
we find that there is at least one water tower in LinkedGeoData that is aligned
with an Educational institution in DBpedia.
The union alignment #5 should have been as straightforward as alignment
#2. Our approach was able to detect a pattern, however, that might have been
overlooked after looking at individual instances. Netherlands from GeoNames,
for example, should be aligned with the country Netherlands from DBpedia.
However we have possible alias names, such as The Netherlands and Kingdom of
Netherlands, as well a possible linkage error to Flag of the Netherlands.svg gener-
ated while importing Wikipedia data into DBpedia (the error seems systematic,
see Jordan in #6).
Alignment #7 was able to explain 8 of the 10 license plate codes in the state
(bundesland) of Saarland6 . The ones that it missed were Ottweiler (OTW) and
the police vehicle codes (SAL). Since the vehicle code SAL is not associated with
any populated places in Saarland, it is quite possible that it does not get men-
tioned in LinkedGeoData. Our approach thus provides a deeper insight into the
nature of the sources. #9 tries to find the composition of the state of New Jer-
sey. 100% of the instances in New Jersey from LinkedGeoData can be accounted
6
http://www.europlates.com/publish/euro-plate-info/german-city-codes
10 Rahul Parundekar, José Luis Ambite, and Craig A. Knoblock
for in the 9 counties. New Jersey actually has 21 counties7 . This suggests that
instances in New Jersey in LinkedGeoData that are linked to DBpedia are not
a complete representation resulting in an equivalent alignment. The quality of
the results generated by our extensional approach are tied to the quality of the
instances in the dataset. We find, however, that such alignments, even though
they might be partially incorrect, give an accurate representation of the actual
instances in the dataset and highlight the practical quality of the links in the
Web of Linked Data.8 Finally, alignment #10 describes how the concept Wa-
terways in LinkedGeoData can be defined as the union concept of Streams and
Rivers in DBpedia. The complete set of alignments discovered by our algorithm
are available on our group page.9
4 Related Work
Ontology alignment has been a well explored area of research since the early
days of ontologies. It has received renewed interest in recent years with the
rise of the Semantic Web. Euzenat & Shvaiko [3] provide a comprehensive dis-
usssion on Ontology Matching approaches. A closely related area of study to
ontology alignment is schema matching. Bernstein et al. [1] summarize the de-
velopments in this field in the past ten years. Though most work done in the
Web of Linked Data is on linking instances across different sources, an increasing
number of authors have looked into aligning the sources ontologies in the past
couple of years. Jain et al. [4] describe the BLOOMS approach which uses a
central forest of concepts derived from topics in Wikipedia. An update to this is
the BLOOMS+ approach [5] that aligns Linked Open Data ontologies with an
upper-level ontology called Proton. Though we employ a simple set subsump-
tion technique to identifying alignments, our use of restriction classes is able to
find a large set of alignments in cases like aligning GeoNames with DBpedia or
Proton, while BLOOMS & BLOOMS+ are unable to find alignments because of
the small number of classes in GeoNames that have vague declarations. Cruz et
al. [2] describe a dynamic ontology mapping approach called AgreementMaker
that uses similarity measures along with a mediator ontology to find mappings
using the labels of the classes. Building ontologies of Linked Data sources using a
statistical method has also been described in Völker et al. [8]. This work induces
schemas for RDF data sources by generating OWL 2 axioms using intermediate
associativity table of instances and concepts (called transaction datasets) and
mining associativity rules from it.
7
http://en.wikipedia.org/wiki/List of counties in New Jersey
8
In [7] we compared the extensional versus intensional perspective on ontology align-
ment. In a nutshell, the extensional alignment gives a precise characterization of
the current relationship between the data in the sources, regardless of the intended
meaning of the concept definitions. For example, a source may define instances as
universities, but linkage can show that it only contains American universities.
9
http://www.isi.edu/integration/data/UnionAlignments
Table 1. Example alignments from the GeoNames and DBpedia datasets, with larger sets from DBpedia and smaller sets from GeoNames
0 A|
# Sub-group {p1 , v1 , p2 } List(v2 ) RU = |U
|UL |
|UA | |UL | Outliers # Explained Instances
1 {rdf:type, S.SCH, S.SCHC, 0.9801 396 404 S.BLDG (3/122), S.EST (1/13), 403
dbpedia:EducationalInstitution, S.UNIV S.LIBR (1/7), S.HSP (1/31),
geonames:featureCode} S.MUS (1/43)
2 {dbpedia:country, ES 0.9997 3917 3918 IT (1/7635) 3918
dbpedia:Spain,
geonames:countryCode}
3 {dbpedia:region, geonames:2989247, 1.0 754 754 754
dbpedia:Basse-Normandie, geonames:2996268,
geonames:parentADM2} geonames:3029094
4 {rdf:type, S.AIRB, S.AIRP 0.9924 1981 1996 S.AIRF (9/22), S.FRMT (1/5), 1996
dbpedia:Airport, S.SCH (1/404), S.STNB (2/5)
geonames:featureCode} S.STNM (1/36), T.HLL (1/61)
Table 2. Example alignments from the DBpedia and GeoNames datasets, with larger sets from GeoNames and smaller sets from DBpedia
0 A|
# Sub-group {p1 , v1 , p2 } List(v2 ) RU = |U
|UL |
|UA | |UL | Outliers # Explained Instances
5 {geonames:countryCode, dbpedia:Netherlands, 0.9802 1939 1978 dbpedia:Kingdom of 1940
NL, dbpedia:The Netherlands, the Netherlands
dbpedia:country} dbpedia:Flag of the
Netherlands.svg
6 {geonames:countryCode, dbpedia:Jordan 0.95 19 20 20
JO, dbpedia:Flag of Jordan.svg
Aligning Unions of Concepts in Ontologies of Geospatial Linked Data
dbpedia:country}
11
Table 3. Example alignments from the LinkedGeoData and DBpedia datasets, with larger sets from DBpedia and smaller sets from
LinkedGeoData
Rahul Parundekar, José Luis Ambite, and Craig A. Knoblock
0 |UA |
# Sub-group {p1 , v1 , p2 } List(v2 ) RU = |UL |
|UA | |UL | Outliers # Explained Instances
7 {dbpedia:bundesland, HOM, IGB, MZG, 0.93 46 49 46
Saarland, NK, SB, SLS,
lgd:OpenGeoDBLicensePlateNumber} VK, WND
8 {rdf:type, lgd:Amenity, lgd:K2543, 0.9901 2609 2610 2609
dbpedia:EducationalInstitution, lgd:School, lgd:University,
rdf:type} lgd:WaterTower
Table 4. Example alignments from the LinkedGeoData and DBpedia datasets, with larger sets from LinkedGeoData and smaller sets
from DBpedia
0 |UA |
# Sub-group {p1 , v1 , p2 } List(v2 ) RU = |UL |
|U A | |UL | Outliers # Explained Instances
9 {lgd:gnisST alpha, Atlantic, Burlington, 1.0 214 214 214
NJ, Cape May, Hudson,
dbpedia:subdivisionName} Hunterdon, Monmoth,
New Jersey, Ocean, Passaic
10 {rdf:type, dbpedia:Stream, 0.97 33 34 dbpedia:Place(1/94989) 34
lgd:Waterway, dbpedia:River
rdf:type
12
Aligning Unions of Concepts in Ontologies of Geospatial Linked Data 13
5 Conclusions and Future Work
We described an approach to identifying union alignments in geospatial data
sources on the Web of Linked Data. By extending our definition of restriction
classes with the disjunction operator, we were able to find alignments of union
concepts from one source to larger concepts from the other source. Our approach
produced union alignments as results that found that concepts at different levels
in the ontologies of two sources can be mapped even when there was no direct
equivalence. We were also able to find outliers that enable us to identify inconsis-
tencies in the instances that are linked by looking at the alignment pattern. The
results provide deeper insight into the nature of the alignments of Geospatial
Linked Data.
Though the scope of this paper is the geospatial domain, our algorithm can
be used in other domains as well. Our next step is to explore other domains like
zoology and genetics for union alignments. Other possible future work is in the
mapping and understanding of the properties in the sources. Our preliminary
findings show that the results of this paper can be used to find patterns in
the properties. For example, the countryCode property in GeoNames is closely
associated with the country property in DBpedia, though their ranges are not
exactly equal. We believe that an in-depth analysis of the alignment of ontologies
of sources is warranted with the recent rise in the links in the Linked Data cloud.
This is an extremely important step for the grand Semantic Web vision.
Acknowledgements
This research is based upon work supported in part by the National Science
Foundation under award number IIS-1117913.
References
1. Bernstein, P., Madhavan, J., Rahm, E.: Generic schema matching, ten years later.
Proceedings of the VLDB Endowment 4(11) (2011)
2. Cruz, I., Palmonari, M., Caimi, F., Stroe, C.: Towards on the go matching of linked
open data ontologies. In: Workshop on Discovering Meaning On The Go in Large
Heterogeneous Data. p. 37 (2011)
3. Euzenat, J., Shvaiko, P.: Ontology matching. Springer-Verlag (2007)
4. Jain, P., Hitzler, P., Sheth, A., Verma, K., Yeh, P.: Ontology alignment for linked
open data. The Semantic Web–ISWC 2010 pp. 402–417 (2010)
5. Jain, P., Yeh, P., Verma, K., Vasquez, R., Damova, M., Hitzler, P., Sheth, A.:
Contextual ontology alignment of lod with an upper ontology: A case study with
proton. The Semantic Web: Research and Applications pp. 80–92 (2011)
6. Parundekar, R., Knoblock, C.A., Ambite, J.L.: Aligning geospatial ontologies on the
linked data web. In: Proceedings of the GIScience Workshop on Linked Spatiotem-
poral Data. Zurich, Switzerland (2010)
7. Parundekar, R., Knoblock, C.A., Ambite, J.L.: Linking and building ontologies of
linked data. In: Proceedings of the 9th International Semantic Web Conference
(ISWC 2010). Shanghai, China (2010)
8. Völker, J., Niepert, M.: Statistical schema induction. The Semantic Web: Research
and Applications pp. 124–138 (2011)