Back to search

Πανεπιστήμιο Μακεδονίας

Ανάπτυξη συστήματος πραγματικού χρόνου για τη προστασία από επιθέσεις κακόβουλης προσέγγισης ανήλικων μέσω διαδικτύου με ασαφή εκτίμηση επικινδυνότητας

Abstract

dc:description

Children and teenagers are exposed to serious risks while they are talking online. Many incidents are reported where children are abused by abuts though Internet based communications. The rising reports of these attacks, which are globally defined as ‘grooming’, increase parents’ concerns about how they can protect their children effectively. In fact, the problem is exacerbated by two main factors: a) the lack of awareness by children concerning the risk they are exposed to while taking online and b) the lack of adequate parental knowledge on how to provide effective protection to children. Consequently, there is a need for developing adequate defenses for children. This thesis presents the research and development Grooming Attack Recognition System (GARS) which aims to detect, assess and control grooming risks in real time by instantly warning designated parents. In this respect, GARS calculates in real time a total “risk value” as the synthesis of four (4) particular risk values: a) the outcome of document classification (DC), b) the outcome of personality recognition (PR), c) the analysis of previously recorded risk values (UH) and d) the exposure time.The four (4) particular risk values are the outcomes of corresponding fuzzy logic controllers. These controllers use as inputs the data which come from the corresponding processes and employ proper fuzzy rules and membership functions calculate the corresponding risk value. The four particular risk values are then properly weighted to give the total risk value. The associated weights reflect the relative importance of each particular value and can be used to adjust the overall system’s sensitivity. This adjustment is necessary based on the fact that GARS should be adapted to the specific needs of each child given that grooming risk varies by age and gender.The total risk value calculated by GARS, reflects the total risk children are exposed to. When the total risk value extends some designated thresholds, a warning signal is sent directly to the parent alerting about the potential threat. GARS operation can be implemented either: a) with all functionalities occurring locally or b) remotely, where clients send dialog parts to a centralized computer which calculates and returns the total risk value. Laboratory evaluations have shown that a remote GARS server can handle a large number of connected clients without a significant strain on resources. Moreover, adjusting the alarm threshold at 0.75 the percentage of incorrect attack recognitions (False Positives) remains low whereas the percentage of False Negatives is high especially for the less sensitive adjustment. However, this issue can be addressed by the adjustment of the corresponding alarm threshold separately for each sensitivity adjustment. Using these settings the False Negative rate can reach zero levels with corresponding False Positives up to 0.2.

Degree

thesis:*
Grantor dc:publisher
Πανεπιστήμιο Μακεδονίας
Year dc:date
2014

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Michalopoulos, Dimitrios
  • Μιχαλόπουλος, Δημήτριος

Subjects

dc:subject × 26

Rights

dc:rights
Statement dc:rights
  • BY_NC_ND
Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/31553

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Michalopoulos, Dimitrios; Μιχαλόπουλος, Δημήτριος. Ανάπτυξη συστήματος πραγματικού χρόνου για τη προστασία από επιθέσεις κακόβουλης προσέγγισης ανήλικων μέσω διαδικτύου με ασαφή εκτίμηση επικινδυνότητας. Πανεπιστήμιο Μακεδονίας, 2014. http://hdl.handle.net/10442/hedi/31553