=Paper=
{{Paper
|id=Vol-2288/om2018_poster1
|storemode=property
|title=Exploiting BabelNet for generating subsumption
|pdfUrl=https://ceur-ws.org/Vol-2288/om2018_poster1.pdf
|volume=Vol-2288
|authors=Mouna Kamel,Daniela Schmidt,Cássia Trojahn,Renata Vieira
|dblpUrl=https://dblp.org/rec/conf/semweb/KamelSTV18
}}
==Exploiting BabelNet for generating subsumption==
https://ceur-ws.org/Vol-2288/om2018_poster1.pdf
Exploiting BabelNet for generating subsumption
Mouna Kamel1 , Daniela Schmidt2 , Cassia Trojahn1 , and Renata Vieira2
1
Institut de Recherche en Informatique de Toulouse, Toulouse, France
{mouna.kamel,cassia.trojahn}@irit.fr
2
Pontificia Universidade Catolica do Rio Grande do Sul, Porto Alegre
daniela.schmidt@acad.pucrs.br, renata.vieira@pucrs.br
1 Introduction
Whereas the ontology matching field has developed fully in the last decades, most
matching approaches are still limited to generating equivalences between entities of
different ontologies. However, for many tasks, finding subsumption relations may be
useful. Despite the variety of matching approaches in the literature, most of them rely
on string-based techniques as an initial estimate of the likelihood that two elements
refer to the same real world phenomenon, hence, the found correspondences represent
equivalences with terms similarly written rather than subsumptions. This paper presents
an approach relying on background knowledge from BabelNet (BN) [3] and on the no-
tion of context. The latter has been exploited in different ways in ontology matching
[2, 4]. They are used for disambiguating the senses that better express the meaning of
ontology concepts when looking for subsumption relations between them in BN.
2 Proposed approach
The matching process is divided in two steps. The first step disambiguates the ontology
concept, and the second looks for a subsumption relation between two concepts.
Concept disambiguation. It finds the semantically closer BN synset for a concept. We
adopt the notion of context as a bag of words. For each ontology concept c, from the
source s and target t ontologies, the context ctxc is constructed from the available infor-
mation about the concept (ID, labels, information on super and sub-concepts, etc.). The
context of BN synsets ctxbn is constructed from their sense and main glosses terms. We
adapt the word sense disambiguation method of Lesk [1], which relies on the calculation
of the word overlap between the sense definitions of two or more target terms. Here, we
overlap the context ctxc and all ctxbn , coming from the synsets retrieved when looking
for c in BN. We retrieve the highest overlap. The overlap function is based on the edit
distance similarity between words rather than on the exact match.
Subsumption detection. Given cs and ct concepts from the source and target ontolo-
gies, and their respectively retrieved synsets syns and synt obtained in the previous
step, we look for a subsumption relation between cs and ct . For that purpose, we
check if synt belongs to the set of hypernyms Hyper(syns ), where Hyper(syns )=
Hyperk (syns ) and k is length of the path from syns to one of its hypernym synsets,
S
k
based on a depth-first search strategy.
2 Mouna Kamel, Daniela Schmidt, Cassia Trojahn, and Renata Vieira
3 Experimentation
Material and methods. We used the set of 7 ontologies from the OAEI conference
data set that are involved in the 21 available reference alignments. In our experiments,
compounds with no entry in BN have been pre-processed by removing the modifiers
(e.g. “Invited speaker” is a “Speaker”). We empirically selected k=2 for the path length
and 0.8 as edit distance threshold. We used as reference the subsumptions inferred from
the available equivalence reference alignments, using Hermit and the Alignment API
4.5. As many concepts do not have any super or sub concepts, we considered 2 settings:
contexts as introduced above and the whole ontology as context for each concept. The
best results, which are reported here, were obtained with the latter.
Results and discussion. Table 1 shows the results (measures were computed using
the Alignment API). Overall, the best results are obtained when considering alignments
close to those expected (extended and semantic measures) rather than exact ones. Look-
ing at the results for each pair of ontologies, the best results where obtained for different
pairs when using the different measures: edas-ekaw (classical), confOf-edas (extended)
and conference-sigkdd (semantic). The overall low results are mainly due to two rea-
sons: a high number of concepts can not be found in BN and using the modifier does
help so much in this task; the construction of contexts suffers from the lack of annota-
tions in the ontologies (as well many concepts do not have any super or sub concepts),
and hence, contexts are not rich enough for disambiguating the synsets.
Table 1. Results for the 21 pairs (and those discarding empty alignments) and best pair results.
Average (21 pairs) Best pair results
Classical Extended Semantic Classical Extended Semantic
Prec Rec Prec Rec Prec Rec Prec Rec Prec Rec Prec Rec
.06 (.23) .02 (.07) .14 (.16) .05 (.06) .02 (.02) .22 (.22) .22 .08 .50 .11 .14 .15
4 Conclusions
We presented an approach for generating subsumption correspondences relying on Ba-
belNet. This task is still a gap in the field and the initial results presented here can be
improved in different ways. We plan to improve the disambiguation strategy, exploit-
ing word embeddings, to automatically enrich the ontology with annotations, to adopt
a hybrid approach combining both lexical and background knowledge, to work on the
confidence of the correspondences, and to look for other relations like meronymy.
References
1. M. Lesk. Automatic sense disambiguation using machine readable dictionaries: How to tell a
pine cone from an ice cream cone. In SIGDOC, pages 24–26. ACM, 1986.
2. A. Maedche and S. Staab. Measuring Similarity between Ontologies. In 13th Conf. on Knowl-
edge Engineering and Knowledge Management, pages 251–263, 2002.
3. R. Navigli and S. P. Ponzetto. BabelNet: The automatic construction, evaluation and applica-
tion of a wide-coverage multilingual semantic network. AI, 193:217–250, 2012.
4. F. C. Schadd and N. Roos. Coupling of Wordnet Entries for Ontology Mapping Using Virtual
Documents. In 7th Workshop on Ontology Matching, pages 25–36, 2012.