Towards Classification of Web Ontologies Using the Horizontal and Vertical Segmentation

Abstract

Abstract. The new era of the Web is known as the semantic Web or the Web of data. The semantic Web depends on ontologies that are seen as one of its pillars. The bigger these ontologies, the greater their exploitation. However, when these ontologies become too big other problems may appear, such as the complexity to charge big files in memory, the time it needs to download such files and especially the time it needs to make reasoning on them. We discuss in this paper approaches for segmenting such big Web ontologies as well as its usefulness. The segmentation method extracts from an existing ontology a segment that represents a layer or a generation in the existing ontology; i.e. a horizontally extraction. The extracted segment should be itself an ontology.Keywords: Ontology; Segmentation; OWL; Semantic Web. IntroductionThe Web ontologies present several interests for the Web, such as annotating data, distinguishing between homonyms and polysemy, generalizing or specializing concepts, driving intelligent user interfaces and even inferring entirely new (implicit) information [3] [4]. Ontologies are created by ontology engineers with the help of domain experts [5]. Let us take as an example an ontology representing a population. This ontology should have information about the citizens, their dates of birth, relationships, hobbies, addresses, competences, jobs, etc. It seems to be a great ontology allowing us to get new information about one person's tendencies, how these tendencies may be affected by his relationships; also, the companies can use it to target their advertisements. But, for one reason or another, we may not be concerned by the population under a certain age, or may be interested only in a particular city's population. For such purposes, we assume that segments of such big ontologies that contain only the desired information will respond better to the users' expectations.Several studies are focused on the extraction, classification [6] [7] and segmentation [3] [8] of data in Web ontology; these data can be represented in the ontology web. For example, N. Gherabi et al [9] present a new approach to mapping data stored in relational databases in the semantic Web, it uses simple mappings based on certain specifications of the database schema and explain how relational databases can be used to define a mapping mechanism between the relational database and the OWL ontology. In another work [10], the authors have developed a method to convert UML schemas to Web Ontology.J. Seidenberg and A. Rector have presented in [3] a method for extracting small segments from large ontologies using the GALEN as an example. The segmentation algorithm they have presented was based on the classes hierarchy, which leads to segmenting the ontology by extracting a specific class hierarchy. Such method responds to segmenting ontologies like the GALEN ontology where one could be interested in a concept like HEART and all its super classes and subclasses.

Semantic data mining in the information age: A systematic review

“…The extracted ontology will consist only of the required knowledge related to the study domain, and will be smaller than the original ontology. For example, Nejjahi et al 94 presented a new ontology segmentation technique based on ontology classes and properties to reduce the size of an ontology model, allowing easier download for users. In addition, Mohsen et al 95 proposed an algorithm that applied power sets to compress the ontology, and the results showed that when redundant concepts were eliminated, the performance of an ontology was slightly affected.…”

Diabetes Prediction Using Machine Learning Algorithms and Ontology

“…After preparing the dataset, we import it into Weka software, which contains tools for data preparation, classification [16], clustering, association rule exploration, visualization [17] and Similarity [18]. We used the six most commonly used classifiers to classify binary datasets (SVM, KNN, ANN, Logistic Regression, Naïve Bayes, Decision Tree).…”

An Enhanced Method for Web Ontologies Encryption

“…Such kind of ontologies include the Gene Ontology (GO) [5], an ontology of the biological area. In [6], the authors used this classification to reduce the size of ontologies by extracting only segments that correspond to a criterion.…”