Abstract
Ontologies are the backbone of the Semantic Web. As a result, the number of existing ontologies and the number of topics covered by them has increased considerably. With this, reusing these ontologies becomes preferable to constructing new ontologies from scratch. However, a user might be interested in a part and/or a set of parts of a given ontology, only. Therefore, ontology modularization, i.e., splitting up an ontology into smaller parts that can be independently used, becomes a necessity. In this paper, we introduce a new approach to partition ontology based on the seeding-based scheme, which is developed and implemented through the Ontology Analysis and Partitioning Tool (OAPT). This tool proceeds according to the following methodology: first, before a candidate ontology is partitioned, OAPT optionally analyzes the input ontology to determine, if this ontology is worth considering using a predefined set of criteria that quantify the semantic and structural richness of the ontology. After that, we apply the seeding-based partitioning algorithm to modularize it into a set of modules. To decide upon a suitable number of modules that will be generated by partitioning the ontology, we provide the user a recommendation based on an information theoretic model selection method. We demonstrate the effectiveness of the OAPT tool and validate the performance of the partitioning approach by conducting an extensive set of experiments. The results prove the quality and the efficiency of the proposed tool.
Highlights
Ontologies are the backbone of the Semantic Web, which provides facilities for integrating, searching, and sharing information on the Web by making it understandable for machines [25,26]
We develop a membership function, MemFun, where each concept is associated with a flag, F, such that if the F of concept c is false, it means c is not assigned to any partition yet and the membership function is called for the concept c
In order to evaluate our proposed system, we looked at two different aspects: First, we ran a number of experiments using a set of ontologies with different characteristics and measured a number of performance metrics
Summary
Ontologies are the backbone of the Semantic Web, which provides facilities for integrating, searching, and sharing information on the Web by making it understandable for machines [25,26]. According to a study by d’Aquin et al [10] already in 2007, at least 7000 ontologies existed in the Semantic Web, providing an unprecedented set of resources. In this situation, in order to create a knowledge base for a specific scenario, the developer has two options: creating proper ontologies from scratch or reusing existing ones. Extracting only relevant parts from ontologies that often contain thousands of concepts is a key challenge. C and P are two disjoint sets of classes (concepts) and properties, respectively. A is a set of axioms and I is a set of instances associated with the set concepts C and properties P
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
