Feature extraction of social media bots for pattern recognition and validation

العناوين الأخرى

استخراج خصائص-مميزات البوتس الخاصة بموقع التواصل الاجتماعية للتعرف على أنماطها و التحقق من صحتها

مقدم أطروحة جامعية

al-Lawzi, Daniyah Abd Zahir

مشرف أطروحة جامعية

al-Qaddumi, Ashraf Ahmad

أعضاء اللجنة

Khasawinah, Baha al-Din
Uqayli, Ibrahim
al-Qassas, Rad

الجامعة

جامعة الأميرة سمية للتكنولوجيا

الكلية

كلية الملك الحسين لعلوم الحوسبة

القسم الأكاديمي

قسم علم الحاسوب

دولة الجامعة

الأردن

الدرجة العلمية

ماجستير

تاريخ الدرجة العلمية

2015

الملخص الإنجليزي

Online social networks attract more and more users every day.

With this huge number of users many attacks such as phishing and automated social engineering attacks emerge.

The latest of these attacks is the use of social media bots for malicious purposes.

This thesis focuses on Facebook as one of the most used Online Social Networks (OSN’s) that offers many opportunities not just for users to communicate, but it also enables business owners, famous figures and politicians to create their pages and keep in touch with their crowds easily.

The misuse of OSN’s will be discussed to show how social media bots are used to draft the public opinion, spread fake reputation for products and services and gather user’s information.

In our work we suggested a Facebook bot detector using machine learning.

We used a Facebook application designed to gather user’s information by asking their permission.

After gathering the user’s information we asked some social media bot providers to provide us with bot accounts to use this application so we can gather their information as well.

The gathered information formed a dataset of bots and humans where we had a total of 759 human and bot accounts.

Features were extracted from the database using python scripts and using Weka different machine learning algorithms were tested.

These algorithms were able to learn the behavior of Facebook bot accounts and compare it to the normal user’s accounts.

The random forest test gave us the best detection ratio that reached 96.31 %.

التخصصات الرئيسية

الهندسة الكهربائية
تكنولوجيا المعلومات وعلم الحاسوب

الموضوعات

• مواقع التواصل الاجتماعي
• الإجراءات الأمنية

عدد الصفحات

57

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-651176