Visible to the public Graph Centrality Based Spam SMS Detection

TitleGraph Centrality Based Spam SMS Detection
Publication TypeConference Paper
Year of Publication2019
AuthorsIshtiaq, Asra, Islam, Muhammad Arshad, Azhar Iqbal, Muhammad, Aleem, Muhammad, Ahmed, Usman
Conference Name2019 16th International Bhurban Conference on Applied Sciences and Technology (IBCAST)
ISBN Number978-1-5386-7729-2
Keywordscentrality scores, classifier, data mining, Data preprocessing, degree centrality, electronic messaging, feature extraction, Graph Centrality, graph centrality measures, graph theory, graphs centrality metrics, Hidden Markov models, Human Behavior, human factors, labeled SMS, Measurement, Metrics, pubcrawl, Scalability, security of data, short messages, short messages usage, social networking (online), spam detection, spam messages, spam SMS detection, status updates, Support vector machines, unclassified SMS, unlabeled SMS, unsolicited e-mail
Abstract

Short messages usage has been tremendously increased such as SMS, tweets and status updates. Due to its popularity and ease of use, many companies use it for advertisement purpose. Hackers also use SMS to defraud users and steal personal information. In this paper, the use of Graphs centrality metrics is proposed for spam SMS detection. The graph centrality measures: degree, closeness, and eccentricity are used for classification of SMS. Graphs for each class are created using labeled SMS and then unlabeled SMS is classified using the centrality scores of the token available in the unclassified SMS. Our results show that highest precision and recall is achieved by using degree centrality. Degree centrality achieved the highest precision i.e. 0.81 and recall i.e., 0.76 for spam messages.

URLhttps://ieeexplore.ieee.org/document/8667174
DOI10.1109/IBCAST.2019.8667174
Citation Keyishtiaq_graph_2019