Yazar "Özel, Selma Ayşe" seçeneğine göre listele

Listeleniyor 1 - 7 / 7
  • [ X ]
    Öğe
    Detection of cyberbullying on social media messages in Turkish
    (Institute of Electrical and Electronics Engineers Inc., 2017) Özel, Selma Ayşe; Akdemir, Seyran; Saraç, Esra; Aksu, Hülya
    The increased use of the Internet and the ease of access to online communities like social media have provided an avenue for cybercrimes. Cyberbullying, which is a kind of cybercrime, is defined as an aggressive, intentional action against a defenseless person by using the Internet, social media, or other electronic contents. Researchers have found that many of the bullying cases have tragically ended in suicides; hence automatic detection of cyberbullying has become important. The aim of this study is to detect cyberbullying on social media messages written in Turkish. To our knowledge, this is the first study which makes cyberbully detection on Turkish texts. We prepare a dataset from Instagram and Twitter messages written in Turkish and then we applied machine learning techniques that are Support Vector Machines (SVM), decision tree (C4.5), Naïve Bayes Multinomial, and k Nearest Neighbors (kNN) classifiers to detect cyberbullying. We also apply information gain and chi-square feature selection methods to improve the accuracy of classifiers. We observe that when both words and emoticons in the text messages are taken into account as features, cyberbully detection improves. Among the classifiers, Naïve Bayes Multinomial is the most successful one in terms both classification accuracy and running time. When feature selection is applied classification accuracy improves up to 84% for the dataset used. © 2017 IEEE.
  • [ X ]
    Öğe
    NESNELERİN İNTERNETİ AĞLARINDA TIKANIKLIK KONTROL MEKANİZMALARI İLE RPL AMAÇ FONKSİYONLARININ KARŞILIKLI PERFORMANS ANALİZİ
    (2022) Söyü, Rıdvan; Demir, Alper Kamil; Özel, Selma Ayşe
    Fiziksel nesnelerin duyargalar ve yazılımlar aracılığıyla sürekli olarak haberleşmesini sağlamak için geliştirilen Nesnelerin İnterneti (IoT) teknolojileri gün geçtikçe hayatımıza daha fazla girmektedir. Ancak, IoT ağları henüz IP ağlarındaki gibi standartlara sahip değildir. Bu ağlar için mevcut protokollerin ne kadar yeterli olduğu ve yeni protokollere ihtiyaç olup olmadığı hala araştırma konusudur. Tıkanıklık kontrolü de bu konulardan biri olup, IoT ağlarındaki sürekli ve yoğun bilgi akışından dolayı büyük öneme sahiptir. IoT ağlarında IP ağlarının aksine UDP tercih edildiğinden tıkanıklık kontrolü uygulama katmanında CoAP tarafından yapılır. Literatürde farklı CoAP tıkanıklık kontrol mekanizmaların karşılaştırıldığı performans analizleri bulunsa da CoAP tıkanıklık kontrol mekanizmaları ile yönlendirme protokolü RPL'in farklı amaç fonksiyonları arasındaki ilişkiyi ve performansa etkisini inceleyen bir çalışma bulunmamaktadır. Bu çalışma kapsamında farklı protokoller kullanılarak tasarlanan ağ yığını kombinasyonları Cooja benzetim ortamında araştırılmıştır. İstemci sayısının 3 ve 9, paket teslim oranı değerinin ise %80, %90 ve %100 olarak alındığı tüm benzetimlerden elde edilen ortalama gecikme ve işlem hacmi metrikleri incelendiğinde Objective Function 0 (OF0)’ın, Minimum Rank of Hysteresis Objective Function (MRHOF) algoritmasına göre daha iyi performans gösterdiğini ve CoCoA Strong'un en iyi performans gösteren tıkanıklık kontrolü mekanizması olduğu görülmüştür.
  • [ X ]
    Öğe
    Similarity detection between Turkish text documents with distance metrics
    (Institute of Electrical and Electronics Engineers Inc., 2017) Kaya, Mümine Keleş; Özel, Selma Ayşe
    The aim of this study is to compare the successes of various distance metrics and to determine the most appropriate methods in order to detect similarities among textual documents written in Turkish. Computing similarities between text documents is the basic step of plagiarism detection, and text mining methods like author detection, text classification and clustering. Therefore, plagiarism detection and text mining applications will be more successful by using the distance metrics that are determined according to the results obtained in this study. For this purpose, chunks of texts in different lengths are selected as the experimental dataset in this study. After that, preprocessing methods are applied to the dataset that is used; therefore new and different experimental scenarios are created by removing stopwords and Turkish characters, and stemming words with Zemberek. According to the experimental results, it is observed that the preprocessing phase increases the accuracy of similarity detection. Especially, stemming using Zemberek increases the success rate. In all cases, the Cosine Similarity method has been observed as more successful than other distance metrics, because of producing more realistic results. © 2017 IEEE.
  • [ X ]
    Öğe
    Think before you share in OSNs: Textual content and connection weight put you at higher privacy risk
    (2022) Çoban, Önder; İnan, Ali; Özel, Selma Ayşe
    The widespread use of OSNs has brought forward the issue of privacy protection over OSNs, as sensitive information of users needs to remain private. Most users are unaware of possible privacy risks associated with sharing personal information in their accounts. Privacy settings of OSNs focus on protecting users' information just by providing them with means of configuring the audience of shared information. As such, privacy risk estimation (or scoring) is a hot topic in the field of OSN research and aims to develop risk measuring tools to ensure user privacy in OSNs. Conventional studies in the area often rely on synthetically generated or survey-based data and do not make any effort to infer private attribute values of users to utilize inference success in privacy scoring of these users. In this study, we propose a novel framework that involves populating a response matrix by using attribute inference and obtaining network aware-risk scores not just by using users' connections but weights of these connections as well. We perform attribute inference of users based on both their textual contents and connections. Our rule-based inference mechanism employed on contents produces inference accuracies ranging from 0.54 to 1.0 depending on the attribute at hand. On the other hand, the inference mechanism involving users' social connections produces inference accuracies of 1.0 almost for all of the considered attributes. We present results and challenges of attribute inference and use inferred attributes in privacy risk scoring. In addition, unlike existing works, we use and show that social tie strengths have to be taken into account in network-aware privacy risk scoring.
  • [ X ]
    Öğe
    Towards the Design and Implementation of an OSN Crawler: A Case of Turkish Facebook Users
    (2020) Çoban, Önder; İnan, Ali; Özel, Selma Ayşe
    Online Social Networks (OSNs) are extremely popular services that allow users to interact with each other and sharecontent. Due to the large amounts of data shared by users, OSNs are also rich data sources for research in social networkanalysis. Studying the usage of OSNs helps to understand users’ content-sharing behavior and privacy concerns. In order to doso, collecting data is a necessary first step. However, Application Programming Interfaces (APIs) provided by OSN providers haveseveral limitations which make it difficult to access secured information. In this paper, we present the design and implementationof an OSN crawler, discuss the challenges of this task and our workarounds towards accessing public OSN data. Moreover, weperform analyses of the collected data to indicate users’ sharing behavior and give a detailed discussion of these analyses from theperspective of individual privacy protection over OSNs. Our crawler overcomes most of the restrictions of OSN APIs and collectsall forms of OSN user interactions as well as every bit of public data posted on an OSN. Most of the existing studies collectOSN data using focused crawlers and therefore are capable of collecting only the desired type of data. Our crawler, on the otherhand, provides a holistic view. On the popular Facebook OSN, our crawler captures user relationships like kinship, friendship andattributes like profile items, events, posts, comments, replies, meta-data of activities (i.e., posting time, location, tagged users etc.).To the best of our knowledge, ours is the most comprehensive OSN data collection effort and also the first study focused on thebehavior of OSN users in Turkey.
  • [ X ]
    Öğe
    Türkçe dokümanlardaki benzerliklerin tespiti için mevcut yazılımların karşılaştırılması ve Türkçe karakter kullanımı ile kök almanın etkisinin incelenmesi
    (2014) Keleş, Mümine Kaya; Özel, Selma Ayşe
    Web ortamındaki bilginin çoğalıp, Internet ve bilgi teknolojilerinin yaygın kullanılması hemen her alanda intihal vakalarının artmasına neden olmuştur. Örneğin, akademik ortamda bazı öğrenciler kendilerine eğitmenleri tarafından verilen ödevler üzerinde çeşitli intihal yöntemlerini uygulamaktadırlar. Bazı öğrenciler başkalarının çalışmasını herhangi bir değişiklik yapmadan ve sahibine atıfta bulunmadan kendi çalışması gibi gösterirken, bazı öğrenciler de diğerlerinin çalışmasını sadece bazı küçük değişiklikler yaparak sunmaktadır. Bu çalışmada amacımız intihal tespit yazılımlarından CopyCatchGold, Sherlock, SIM, WCopyFind, JPlag, YTÜ Kemik Grubu tarafından hazırlanan Metin Eşleştirme Sistemi ve Doküman Benzerliği programları ile kendi kodladığımız Kosinüs, Dice ve Jaccard metin benzerlik ölçütlerinin Türkçe örnek veri kümeleri üzerinde performanslarını karşılaştırmaktır. Buna ek olarak Türkçe karakter ve kelime köklerinin kullanımının intihal tespiti üzerindeki etkisi incelenmiştir. Sonuç olarak, Türkçe karakter kullanımının benzerlik tespitini azalttığı, ancak kelime köklerinin kullanımının ise intihal tespit araçlarının performansını arttırdığı gözlenmiştir.
  • [ X ]
    Öğe
    Your Username Can Give You Away: Matching Turkish OSN Users with Usernames
    (2021) Çoban, Önder; İnan, Ali; Özel, Selma Ayşe
    Abstract—User profile matching (i.e., user cross-referencing, user identification) aims to find accounts that belong to the same users over different websites or online social networks (OSNs). Solving this problem can be useful for many operations and functionalities such as friend recommendation and link prediction across different OSNs. Additionally, identifying users across different OSNs may enable an adversary to aggregate incomplete information of users. Hereby, an adversary can extract and use online footprint of users to violate their privacy and security via putting them into threats such as identity theft, online stalking, and blackmailing among many others. Usernames are indispensable elements of all websites that require user registration. Even though usernames are generally short strings, they potentially reflect users’ characteristics and habits such as the political sense of belonging, hometown, and so on. In this study, we make an effort to match users of distinct OSNs relying only on their usernames. We use two different approaches based on machine learning and vector-based username similarity to build our learning function. We also explore different feature spaces from the literature and further investigate which approach produces better results. We conducted our experiments on a real-world username data set that is extracted from the OSN accounts of Turkish users we crawled in our previous work. Our results show that building learning function by binary classification outperforms the similarity approach and it achieves the best F-score of 0.921 without feature selection and extension.