Ontology matching is a critical task to realize the Semantic Web vision, by enabling interoperability between ontologies. However, handling large ontologies e ciently is a challenge, given that ontology matching is a problem of quadratic complexity. AgreementMakerLight (AML) is a scalable automated ontology matching system developed to tackle large ontology matching problems, particularly for the life sciences domain. Its new 2.0 release includes several novel features, including an innovative algorithm for automatic selection of background knowledge sources, and an updated repair algorithm that is both more complete and more e cient. AML is an open source system, and is available through GitHub 1 both for developers (as an Eclipse project) and end-users (as a runnable Jar with a graphical user interface).
Ontology matching is the task of nding correspondences (or mappings) between
semantically related concepts of two ontologies, so as to generate an alignment
that enables integration and interoperability between those ontologies [
At its base, ontology matching is a problem of quadratic complexity as it entails comparing all concepts of one ontology with all concepts of the other. Early ontology matching systems were not overly concerned with scalability, as the matching problems they tackled were relatively small. But with the increasing interest in matching large (biomedical) ontologies, scalability became a critical aspect, and as a result, traditional all-versus-all ontology matching strategies are giving way to more e cient anchor-based strategies (which have linear time complexity).
OWL
AgreementMakerLight
RDF
AgreementMakerLight (AML) is a scalable automated ontology matching system
developed to tackle large ontology matching problems, and focused in
particular on the biomedical domain. It is derived from AgreementMaker, one of the
leading rst generation ontology matching systems [
The AML ontology matching framework is represented in Figure 1. It is divided into four main modules: ontology loading, primary matching, secondary matching, and alignment selection and repair
The ontology loading module is responsible for reading ontologies and parsing
their information into the AML ontology data structures, which were conceived
to enable anchor-based matching [
The primary and secondary matching modules contain AML's ontology
matching algorithms, or matchers, with the di erence between them being their time
complexity. Primary matchers have O(n) time complexity and therefore can be
employed globally in all matching problems, whereas secondary matchers have
O(n2) time complexity and thus can only be applied locally in large problems.
The use of background knowledge in primary matchers is a key feature in AML,
and it includes an innovative automated background knowledge selection
algorithm [
The alignment selection and repair module ensures that the nal alignment has
the desired cardinality and that it is coherent (i.e., does not lead to the
violation of restrictions of the ontologies) which is important for several applications.
AML's approximate alignment repair algorithm features a modularization step
which identi es the minimal set of classes that need to be analyzed for coherence,
thus greatly reducing the scale of the repair problem [
The GUI was a recent addition to AML, as we sought to make our system
available to a wider range of users. The main challenge in designing the GUI was
nding a way to visualize an alignment between ontologies that was both scalable
and useful for the user. Our solution was to visualize only the neighborhood of
one mapping at a time, while providing several options for navigating through
the alignment [
AML 1.0 achieved top results in the 2013 edition of the Ontology Alignment
Evaluation Initiative (OAEI) [
AML 2.0 is more e ective than its predecessor (thanks to the improved handling of background knowledge, the richer data structures and the addition of new matching algorithms) without sacri cing e ciency, so we expect it to perform even better in this year's edition of the OAEI.
DF, CP, ES and FMC were funded by the Portuguese FCT through the SOMER project (PTDC/EIA-EIA/119119/2010) and the LASIGE Strategic Project (PEst-OE/EEI/UI0408/2014). The research of IFC was partially supported by NSF Awards CCF{1331800, IIS{1213013, IIS{1143926, and IIS{0812258 and by a UIC-IPCE Civic Engagement Research Fund Award.