of 24 lecture 11: ontology – mediation, merging & aligning

of 24 ece 627, winter ‘132 ontology … mediation, merging and aligning ontologies are shared specifications and can be used for the annotation of multiple data sources (web pages, XML documents, relational databases) however, it cannot be expected that all individuals and organizations will ever agree on using one common terminology or ontology

of 24 ece 627, winter ‘133 ontology … mediation, merging and aligning ontology mediation enables reuse of data across applications on Semantic Web, and sharing of data between heterogeneous knowledge bases principle kinds of mediation: mapping and merging

of 24 ece 627, winter ‘134 ontology … mediation, merging and aligning ontology mapping: correspondences are used to, for example, transforming data between different representations correspondences between ontologies are stored separately – are not part of the ontologies themselves the (semi-)automated discovery of such correspondences is called ontology alignment (mapping: slides 11-15, alignment: 16-22)

of 24 ece 627, winter ‘135 ontology … mediation, merging and aligning ontology merging: a new ontology is created which is the union of the source ontologies – it captures all the knowledge from the original ones the challenge – all correspondences and differences are reflected in the merged ontology (also slide 23)

of 24 ece 627, winter ‘136 ontology mediation important issues - mismatches between concepts, relations, and instances in different ontologies - localization and specification of overlaps

of 24 ece 627, winter ‘137 ontology mediation ontology mismatches two types of mismatches: - conceptualizaiton mismatches - explication mismatches

of 24 ece 627, winter ‘138 ontology mediation ontology mismatches: conceptualization scope mismatch: occurs when two classes have some overlap in their sets of instances, but the sets are not exactly the same model coverage and granularity mismatches: - differences in the part of the domain that is covered by both ontologies - differences in the level of details

of 24 ece 627, winter ‘139 ontology mediation ontology mismatches: explication style of modeling mismatch: the paradigm used to specify a certain concept (for example time) is different the way concepts are describe differs (subclasses vs. attributes) terminological mismatch: concepts are equivalent but represented using different names

of 24 ece 627, winter ‘1310 ontology mediation ontology mismatches: explication encoding mismatch: ontologies are encoded in a different way (using km vs miles for a distance)

of 24 ece 627, winter ‘1311 ontology mediation ontology mapping a specification of the semantic overlap between two ontologies the correspondences between different entities of the two ontologies are expressed using some axioms formulated in a specific mapping language

of 24 ece 627, winter ‘1312 ontology mediation ontology mapping three phases: - mapping discovery - mapping representation - mapping exploitation/execution

of 24 ece 627, winter ‘1313 ontology mediation ontology mapping MAFRA (Mapping FRAmework) - lift and normalization (lifting the content of the ontologies to RDFS and normalization of their vocabularies) - similarity (computation of similarities between ontology entities) - semantic bridging (establishing correspondences between similar entities, in the form of so-called semantic bridges) - execution (exploiting the bridges/mappings for individual transformation) - post-processing (revisiting the mapping specification for improvements)

of 24 ece 627, winter ‘1314 ontology mediation ontology mapping semantic bridge (five aspects): - entity aspect (entities related by a bridge – may be concepts, relations, attributes) - cardinality aspect (a number of ontology entities at both sides of the bridge – 1:n or m:1) - structural aspect (a single bridge may be combined into a more complex bridge) - transformation aspect (how individuals are transformed by associated transformation functions) - constraint aspect (conditions upon whose fulfillment the bridge evaluation depends)

of 24 ece 627, winter ‘1315 ontology mediation ontology mapping a common tendency is the existence of an ontology of mappings that contains the vocabulary for the representation of mappings semantic bridges are captured in the Semantic Bridging Ontology (SBO)

of 24 ece 627, winter ‘1316 ontology mediation ontology alignment it is a process of discovering similarities between two source ontologies input: a number of ontologies output: a specification of the correspondences between ontologies

of 24 ece 627, winter ‘1317 ontology mediation ontology alignment algorithms that perform matching can be divided based on: - a schema-based matching - an individual-based (instant-based) matching or - an element-level matching - a structure-level matching

of 24 ece 627, winter ‘1318 ontology mediation ontology alignment a schema-based matching takes different aspects of the concepts and relations in the ontologies and uses some similarity measure to determine correspondence

of 24 ece 627, winter ‘1319 ontology mediation ontology alignment an individual-based matching takes the individuals which belong to the concepts in different ontologies and compares them to discover similarity between the concepts

of 24 ece 627, winter ‘1320 ontology mediation ontology alignment an element-level matching compares properties of the particular concept or relation, such as a name, and uses those to find similarities

of 24 ece 627, winter ‘1321 ontology mediation ontology alignment a structure-level matching compares the structure of ontologies to find similarities

of 24 ece 627, winter ‘1322 ontology mediation ontology alignment - examples a system takes as input an initial list of similarities between concepts, and additional similarities are found based on the initial similarities and the structure of the ontologies a system takes two pairs of related terms and analyzes the elements which are included in the part that connect the elements a system takes two graph-like structures and produces a mapping between elements of the two graphs that semantically correspond to each other

of 24 ece 627, winter ‘1323 ontology mediation ontology merging is a process of creation of one ontology from two or more source ontologies – it will unify and in general replace the original ontologies approaches: - creating of a new, merged ontology (next slide) - creating a “view” – so called a bridge ontology – it specifies correspondences between entities (slide 13)

of 24 ece 627, winter ‘1324 ontology mediation ontology merging – an example identification of merge candidates based on class-name similarities (the results are presented to the user as a list) selection, by the user, of suggested operations (merge classes, merge properties, …) from the list execution of the requested action (with automatic execution of additional changes) creation of a new list of suggested actions (based on the new structure of the ontology), determination conflicts introduced by the last action, finding possible solutions to these conflicts