Ontology Integration Systems and Tools 46

Ontology integration is a complicated process. It is difficult to find the terms that need to be aligned, and the consequences of a specific mapping (unforeseen implications) are difficult to see. Semi-automatic tools are required to guide the user through the process and focus this attention on the likely points for action, and enable reusability of alignments in the context of ontology maintenance.

A number of ontology integration systems exists that support users to find inter-ontology relationships. Some of these systems can also perform merging and

create a new ontology based on the source ontologies and the alignment relationships. [McGuiness, et al., 2000] provides the first tool to help in the merge process.

(1) A General Framework

Lambrix et al. proposed a general framework for ontology alignment [Lambrix and Tan, 2006], as depicted in the following Figure 2-9. Many ontology alignment systems can be described as instantiations of this framework.

Figure 2-9. A general framework for ontology alignment [Lambrix and Tan, 2006].

In this framework, an alignment algorithm receives two source ontologies as input. The algorithm can include several matchers. These matchers calculate similarities between the terms from the different source ontologies. The matchers can implement strategies based on linguistic matching, structure-based strategies, constraint-based approaches, instance-based strategies, and strategies that use auxiliary information or a combination of these. Alignment suggestions are then determined by combining and filtering the results generated by one or more matchers. The pairs of terms with a similarity value above a certain threshold are retained as alignment suggestions. By using different matchers and combining them and filtering in different ways, different

alignment strategies will be obtained. The suggestions are then presented to the user who accepts or rejects them. The acceptance or rejection of a suggestion may influence further suggestions. Further, a conflict checker is used to avoid conflicts introduced by the alignment relationships. The output of the alignment algorithm is a set of alignment relationships between terms from the source ontologies.

In this framework the matchers use different strategies to calculate similarities between the terms from different source ontologies. They use different kinds of knowledge that is exploited during the alignment process to enhance their effectiveness and efficiency. Some of the approaches employed are described as follows:

z Strategies based on linguistic matching. These approaches make use of textual descriptions of the concepts and relations such as names, synonyms and definitions. The similarity measure between concepts is based on comparisons of the textual descriptions.

z Structure-based strategies. These approaches use the structure of the ontologies to provide suggestions. The similarity of concepts is based on their environment. For instance, using the is-a relation, an environment can be defined using the parents (or ancestors) and the children (or descendants) of a concept.

z Constraint-based approaches. In this case axioms are used to provide

suggestions. For example, knowing that the range and domain of two relations are the same may be an indication that there is a relationship between the relations.

z Instance-based strategies. In some cases instances are available directly or can be obtained. When instances are available, they may be used in defining similarities between concepts.

z Use of auxiliary information. Dictionaries and thesauri representing general or domain knowledge, or intermediate ontologies may be used to enhance the alignment process. They provide external resources to interpret the intended meaning of the concepts and relations in an ontology.

z Combining different approaches. The different approaches use different

strategies to compute similarity between concepts. Therefore, a combined approach may give better results.

(2) SAMBO

SAMBO [Lambrix and Tan, 2006] is an ontology alignment and merging tool developed according to the above framework. SAMBO supports ontologies in the OWL 10 format. The system separates the process into two steps: aligning relations and aligning concepts. In the suggestion mode several kinds of matchers can be used and combined. The pairs of terms with a similarity value above a threshold are shown to the user as alignment suggestions. For each of the alignment suggestions the user can decide whether the terms are equivalent, whether there is an is-a relation between the terms, or whether the suggestion should be rejected. If the user decides that the terms are equivalent, a new name for the term can be given as well. If the user rejects a suggestion where two different terms have the same name, he is required to rename at least one of the terms. At each point during the alignment process the user can view the ontologies represented in trees with the information on which actions have been performed, and the user can check how many suggestions still need to be processed. In addition to the suggestion mode, the system also has a manual mode in which the user can view the ontologies and manually align terms. The source ontologies are illustrated using is-a and part-of hierarchies. The user can choose terms from the ontologies and then specify an alignment operation. After the user accomplishes the alignment process, the system receives the final alignment list and can be asked to

create the new ontology. The system merges the terms in the alignment list, computes the consequences, makes the additional changes that follow from the operations, and finally copies the other terms to the new ontology.

(3) Protege PROMPT

Protege is a tool for creating, editing, browsing, and maintaining ontologies 11. PROMPT is one of its plug-ins, including several interactive tools for ontology merging and aligning [Noy and Musen, 2003]. iPROMPT is the ontology merging tool in the PROMPT suite [Noy and Musen, 2000]. When merging two ontologies, iPROMPT creates a list of initial suggestions based on the underlying alignment algorithms. The suggestions can, for instance, be to merge two terms, or to copy a term to the new ontology. The user can then perform an operation by accepting one of the suggestions or creating his own suggestions. iPROMPT then performs the operation and additional changes that follow from that operation. The list of suggestions is then updated and a list of conflicts and possible solutions to these conflicts is created. This is repeated until the new ontology is ready.

(4) Ontolingua Server

Ontolingua Server is an ontology development environment for collaborative ontology construction, addressing the problem of ontology integration [Farquhua, et al., 1995 and Farquhua, et al., 1997]. This tool allows collaborative ontology building and also provides an ontology library, where tested ontologies are gathered and made publicly available. To allow reuse of the ontologies available at the Ontolingua Server library, a set of integration operations was identified, specified, defined, and made available to ontology builders. Users are allowed three operations: inclusion, polymorphic refinement and restriction (specialization). Inclusion is used when the ontology is included (from the library of ontologies kept by the tool) and used as it is. Polymorphic refinement extends one operation so that it can be used with several

kinds of arguments. Restriction makes simplifying assumptions that restrict the included axioms. The Ontolingua Server also provides facilities for local symbol renaming. This facility enables ontology developers to refer to symbols from other ontologies using names that are more appropriate to a given ontology and to specify how naming conflicts among symbols from multiple ontologies are to be resolved. (5) FOAM

FOAM 12 is a semi-automatic tool for aligning and merging two or more OWL

ontologies. When merging ontologies in semi-automatic mode, FOAM proposes alignment suggestions and the user can accept or reject these suggestions. The output of the system after processing all the suggestions is the accepted list of alignments.