Main Page
Deanship
The Dean
Dean's Word
Curriculum Vitae
Contact the Dean
Vision and Mission
Organizational Structure
Vice- Deanship
Vice- Dean
KAU Graduate Studies
Research Services & Courses
Research Services Unit
Important Research for Society
Deanship's Services
FAQs
Research
Staff Directory
Files
Favorite Websites
Deanship Access Map
Graduate Studies Awards
Deanship's Staff
Staff Directory
Files
Researches
Contact us
عربي
English
About
Admission
Academic
Research and Innovations
University Life
E-Services
Search
Deanship of Graduate Studies
Document Details
Document Type
:
Thesis
Document Title
:
IMPROVEMENT OF TA-ARM ALGORITHM FOR CONCEPT MAP BASED ON TEXT ANALYSIS
تحسين خوارزمية TA-ARM لخريطة المفاهيم بناءً على تحليل النص
Subject
:
Faculty of Computing and Information Technology
Document Language
:
Arabic
Abstract
:
Text analysis is one of the emerged methods to deal with unstructured text data that allows to automatically extract and classify information from text. It has many applications in Knowledge Visualization (KV) field as it severs as the base of transforming textual data into visual presentations such as sketches and conceptual diagrams. Concept map is a form of conceptual diagrams that shows the main concepts of a given text and shows the relationships among these concepts. However, constructing a manual concept map from a given text is deemed to be a tedious task. It is time-consuming, and it demands an extensive effort in reading the textual content and reasoning the relationships among concepts. For this reason, many studies have utilized text analysis methods to develop intelligent algorithms in order to generate concept maps automatically with a high accuracy at a minimal effort. This thesis aims at improving the Text Analysis phase in Text Analysis and Association Rules Mining (TA-ARM) algorithm that automatically generates concept maps from test questions and answer records. The improvement of test questions classification in Text Analysis phase is supposed to lead to more accurately classified test questions into concepts. The distance-weighted K-Nearest Neighbors classifier is applied to classify test questions instead of the traditional K-NN. This approach is expected to handle the imbalanced dataset and increases classification accuracy. A comparison between the distance-weighted K-NN classifier and the traditional K-NN were undertaken. The results have shown better classification performance by the distance-weighted K-NN than by the traditional K-NN in terms of accuracy, precision, recall, and F1-score metrics. the distance-weighted K-NN achieved an accuracy of 95% and F1-score of 96% where the traditional K-NN achieved an accuracy of 71% and F1-score of 55%. These results have confirmed the effectiveness of the proposed improvement.
Supervisor
:
Dr. Salha Binti Abdullah
Thesis Type
:
Master Thesis
Publishing Year
:
1441 AH
2020 AD
Added Date
:
Sunday, June 14, 2020
Researchers
Researcher Name (Arabic)
Researcher Name (English)
Researcher Type
Dr Grade
Email
سارة صالح العمري
Alomari, Sara Saleh
Researcher
Master
Files
File Name
Type
Description
46387.pdf
pdf
Back To Researches Page