Towards a complex alignment evaluation dataset

    Élodie Thiéblin, Ollivier Haemmerlé, Nathalie Hernandez, Cassia Trojahn

           IRIT & Université de Toulouse 2 Jean Jaurès, Toulouse, France
                         {firstname.lastname}@irit.fr

    Keywords: complex alignments, evaluation dataset, complex dataset


1    Motivation and background
Simple ontology alignments, largely studied, link one entity from a source ontol-
ogy to one entity of a target ontology. One of the limitations of these alignments
is, however, their lack of expressiveness which can be overcome by complex align-
ments. Different approaches for generating complex alignments have emerged in
the literature [4,5,6]. However, there is a lack of datasets on which they can be
evaluated.
     Ontology matching is the process of generating an alignment. An alignment
A between a source o1 and a target o2 ontologies is a set of correspondences [2].
Each correspondence is a triple heo1 , eo2 , ri. eo1 and eo2 are the members of the
correspondence: they can be single ontology entities or constructions of these
entities using constructors or transformation functions. r is a relation (e.g., ≡,
≤, ≥) between eo1 and eo2 . We consider two types of correspondences:
  – simple correspondence when both eo1 and eo2 are single entities: e.g. ∀x,
     o1:Person(x) ≡ o2:Human(x) is a simple correspondence.
  – complex correspondence when at least one of eo1 or eo2 is a construction of
     entities, i.e. involving at least a constructor or a transformation function. For
     example, ∀x,y, o1:priceInDollars(x,y) ≡ ∃y1, o2:priceInEuro(x,conversion(y))
     is a complex correspondence with a transformation function (conversion
     that states that y1 = changeRate × y). ∀x, o1:AcceptedPaper(x) ≡ ∃y,
     o2:Paper(x) ∧ o2:acceptedBy(x,y) is a complex correspondence with con-
     structors.
A complex alignment contains at least one complex correspondence.


2    The evaluation dataset
The proposed dataset is based on the OntoFarm dataset [9] composed of 16 on-
tologies on the conference organisation domain and simple reference alignments
between 7 of these ontologies. This dataset has been widely used in the ontology
alignment evaluation domain [8]. The dataset proposed here is a first version of
an extension of the OntoFarm dataset including complex correspondences. 3 out
of the 7 ontologies of the reference alignments have been manually aligned (cmt,
conference and edas), resulting in 3 alignments: cmt-conference, cmt-edas and
conference-edas. The methodology applied to create the complex dataset consists
in manually finding an equivalent construction of target entities for each source
entity. All correspondences have a single entity member and an other member
that is either a single entity (simple correspondence) or a construction (complex
correspondence). The correspondences are diverse for they can be classified with
8 different correspondence patterns or compositions of them [7]. In the 3 align-
ments, the dataset contains 51 complex correspondences. The alignments are
expressed in First Order Logic and in EDOAL1 . The resulting alignments were
translated into OWL axioms as an ontology merging process. The HermiT rea-
soner [3] was used to check the consistency of the merged ontology. The dataset is
available online at http://doi.org/10.6084/m9.figshare.4986368.v4 under
a CC-BY License.


3       Conclusion and future work
We have proposed a complex coherent dataset with complex correspondences
between 3 ontologies of the OntoFarm dataset. As perspectives, the dataset will
be extended with other ontologies of this dataset. The confidence of a correspon-
dence (a value associated with a correspondence to express its confidence degree)
could be added to the dataset. This could express, as in [1], the consensus level
of experts on each correspondence. Finally, we aim at using this dataset for the
purpose of evaluating complex matchers.


References
1. Cheatham, M., Hitzler, P.: Conference v2. 0: An uncertain version of the OAEI
   Conference benchmark. In: ISWC. pp. 33–48. Springer (2014)
2. Euzenat, J., Shvaiko, P.: Ontology Matching. Springer Berlin Heidelberg (2013)
3. Glimm, B., Horrocks, I., Motik, B., Stoilos, G., Wang, Z.: HermiT: An OWL 2
   reasoner. Journal of Automated Reasoning 53(3), 245–269 (2014)
4. Jiang, S., Lowd, D., Kafle, S., Dou, D.: Ontology matching with knowledge rules.
   In: Transactions on Large-Scale Data-and Knowledge-Centered Systems XXVIII,
   pp. 75–95. Springer (2016)
5. Parundekar, R., Knoblock, C.A., Ambite, J.L.: Linking and building ontologies of
   linked data. In: ISWC. pp. 598–614. Springer (2010)
6. Ritze, D., Meilicke, C., Šváb Zamazal, O., Stuckenschmidt, H.: A pattern-based
   ontology matching approach for detecting complex correspondences. In: 4th ISWC
   workshop on ontology matching. pp. 25–36 (2009)
7. Scharffe, F.: Correspondence Patterns Representation. Ph.D. thesis, Faculty of
   Mathematics, Computer Science and University of Innsbruck (2009)
8. Zamazal, O., Svátek, V.: The Ten-Year OntoFarm and its Fertilization within the
   Onto-Sphere. Web Semantics: Science, Services and Agents on the World Wide Web
   43, 46–53 (Mar 2017)
9. Šváb, O., Svátek, V., Berka, P., Rak, D., Tomášek, P.: Ontofarm: Towards an ex-
   perimental collection of parallel ontologies. Poster Track of ISWC 2005 (2005)

    1
        http://alignapi.gforge.inria.fr/edoal.html