Arabic text classification based on data mining techniques

مقدم أطروحة جامعية

al-Jamal, Muhannad Said

مشرف أطروحة جامعية

Hattab, Izz al-Din Shakir Hasan

أعضاء اللجنة

al-Shaykh, Asim A. R.
al-Sarayirah, Bashshar
al-Lahham, Muhammad Ismail Abd al-Rasul

الجامعة

الأكاديمية العربية للعلوم المالية و المصرفية

الكلية

كلية نظم و تكنولوجيا المعلومات

القسم الأكاديمي

قسم نظم المعلومات الحاسوبية

دولة الجامعة

الأردن

الدرجة العلمية

دكتوراه

تاريخ الدرجة العلمية

2012

الملخص الإنجليزي

Classification using association rule mining also known as associative classification (AC) is a promising approach that usually builds higher accurate classifier than classic classification approaches in data mining, including probabilistic, decision trees and covering.

In this thesis, we study the problem of using AC mining in structured English and unstructured Arabic textual data collections to discover extract moderate size classifiers that end-user can easily control and understand.

The outcome is a new AC algorithm that efficiently finds knowledge (class association rules–CARs) from different classification benchmark problems, together with extensive experimental results.

Precisely, we introduce the―Reduced Classification using Association‖ (RCUA) which contains three new procedures : (1) A rule discovery procedure that uses new data representation method to efficiently discover the CARs (rules) in one data scan.

(2) A classifier building procedure that reduces the size of the classifiers and over fitting by generating a moderate size classification systems where end user can easily maintain and interpret (3) An evaluation procedure that instead of utilizing one rule for class assignment during prediction like CBA it bases its class assignment decision on multiple rules and thus improving the classification accuracy of the resulting classifiers.

During extensive experimentations using a range of real world application data (Arabic and English), we revealed that our algorithm is often more accurate than well-known traditional classification techniques and competitive with popular associative classification algorithms.

التخصصات الرئيسية

تكنولوجيا المعلومات وعلم الحاسوب

الموضوعات

• معالجة البيانات
• اللغة العربية
• التنقيب في البيانات

عدد الصفحات

131

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-306664