Enhancing the ontologies matching in semantic web using artificial neural network

العناوين الأخرى

تحسين استجابة التجميعات الموجودة في الويب الدلالي عن طريق استخدام الشبكة العصبية الصناعية

مقدم أطروحة جامعية

Darwish, Rami Hilmi

مشرف أطروحة جامعية

Ghanimat, Rawan

الجامعة

جامعة الأميرة سمية للتكنولوجيا

الكلية

كلية الملك الحسين لعلوم الحوسبة

دولة الجامعة

الأردن

الدرجة العلمية

ماجستير

تاريخ الدرجة العلمية

2018

الملخص الإنجليزي

Ontology matching process aims to find correspondence matching between similar terms in different ontologies related to the same domain.

The ontologies are the main component in the semantic web that provides a new framework consisting of many ontologies, for whose support specific languages were designed such as RDF, OWL, and XML.

The process of finding the relationships between multiple ontologies can be achieved manually, by an error-prone and tedious process, or automatically using rule-based techniques (e.g.

WordNet) and specific equations like edit distance, or learning-based techniques such as machine learning tools (e.g.

artificial neural network, ANN).

This study conducts automatic matching between ontologies.

Five large overlapping ontologies were selected: EKAW, CONFIOUS, CMT, COCUS, and CONFOF.

These ontologies are analyzed based on their constituent formulas and expressions, with schema and instance information used as final RDF-triplets datasets.

The final datasets are encoded manually using synset offset value from WordNet.

Finally, ANN is taken to find correspondence matching between the five ontologies.

Three training algorithms are used to train the ANN separately: gradient descent, conjugate gradient, and Levenberg-Marquardt.

The contribution of our approach is to retrieve more than output at the same time, which means that the ANN has to be trained enough to distinguish that a certain triplet is existing in more than one ontology at the same time.

This will be useful in the future if one of the ontologies is down or under maintenance, offering other choices to end-users to retrieve requested information.

التخصصات الرئيسية

تكنولوجيا المعلومات وعلم الحاسوب

الموضوعات

• الويب الدلالي
• الشبكات العصبية(الحاسبات الإلكترونية)

عدد الصفحات

127

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-795726