XMap++: results for OAEI 2014

In this paper, we present the results obtained by our ontology matching system XMap++ within the OAEI 2014 campaign. XMap++ is a scalable ontology alignment tools capable of matching large scale ontology. This is our second participation in the OAEI, and we can see an overall improvement on nearly every task. 1 State, purpose, general statement XMap (eXtensible Mapping) is an ontology alignment tool for the alignment of OWL entities (i.e., classes, object properties and data properties). XMap++ approach uses different similarity measures of different categories such as string, linguistic, and structural based similarity measures to understand ontologies semantics. A weights vector must, therefore, be assigned to these similarity measures, if a more accurate and meaningful alignment result is favored. Combining multiple measures into a single similarity metric has been solved using weights determined by intelligent strategies [3]. The major drawback of our two previous versions XMapGen and XMapSig [2], despite the fact that they achieved fair results and the aim of their development is to deliver a stable version, the time performance was very low time, especially for the Large Biomedical Ontologies tracks, inability to recognize multiple labels to a single entity as synonyms and inability to recognize labels translated in different languages (e.g Chinese, Czech, Dutch, French, German). After carefully studying this issue, we realize that our algorithm needs more assessment in its performance. This inspires us to consider new strategies in the new version of XMap++ 2014, such as : 1) Using cosine similarity as a string similarity methods to compare the concepts textual descriptions associated with the nodes (labels, names, identity, etc) of each ontology; 2) Involving particular parallel matching on multiple cores or machines for dealing with the scalability issue on ontology matching; 3) Translating labels with different languages using Bing Translator (not use any services which require payment); 4) Interfacing with the Wordnet electronic dictionary using Java Wordnet Interface (JWI) as a Java library. Meanwhile, XMap++ loads WordNet dictionary fully into memory to gain time when it aligns large-scale ontologies. Consequently, the new version XMap++ 2014 has improved both the matching quality and time performance in large scale ontology matching tasks. 1.1 Specific techniques used The workflow and the main components of the system can be seen in the Fig. 1. The XMap++ consists of the following components: Fig. 1. Sketch of Architecture for XMAP++. 1. Matching inputs are two ontologies, source O and target O ′ parsed by an Ontology Parser component; 2. The String Matcher based on linguistic matching compares the textual descriptions of the concepts associated with the nodes (labels, names) of each ontology; 3. The Linguistic matcher jointly aims at identifying words in the input strings, relaying on WordNet [7]. These matching techniques may provide incorrect match candidates, structural matching is used to correcting such match candidates based on their structural context. In order to deal with lexical ambiguity, we introduce the notion of the scope belonging to a concept which represents the context where it is placed [1]. The value of linguistic methods is added to the linguistic matcher or the structure matcher in order to enhance the semantic ambiguity during the comparison process of entity names; 4. The structural matcher aligns nodes based on their adjacency relationships. The relationships (e.g., subClassOf and is-a) that are frequently used in the ontology serve, at one hand, as the foundation of the structural matching; 5. The three matchers perform similarity computation in which each entity of the source ontology is compared with all the entities of the target ontology, thus producing three similarity matrices, which contain a value for each pair of entities. After that, an aggregation operator is used to combine multiple similarity matrices computed by different matchers to a single aggregated similarity matrix. We refer to [3] for more detail about the pruning and splitting techniques on data matrices for two couple of entities; 6. XMap++ uses three types of aggregation operator; these strategies are aggregation, selection and combination [3]; 7. Finally, these values are filtered using a selection according to a defined threshold and the desired cardinality. In our algorithm, we adopt the 1-1 cardinality to find the optimal solution in polynomial time.