Polychronicity tendency-based online behavioral signature

• Main Authors: Ikuesan, Adeyemi Richard, Abd. Razak, Shukor, Venter, Hein S., Salleh, Mazleena
• Format: Article
• Published: Springer Verlag 2019
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://eprints.utm.my/id/eprint/88716/; http://dx.doi.org/10.1007/s13042-017-0748-7

The proliferation of ubiquitous and pervasive computing devices has led to the emergence of research areas like Internet of things, and the Big-Data, which has seen a rise in obfuscation of online identity thus fueling an increase in online anonymity. Online anonymity constitutes a major platform for the exploitation of the potentials of cyber-crime; at the same time, it also inhibits the potential economic power that can be harnessed from the surging Internet population. Methods of online identification, such as usage profiling, demographic profiling, cookie-based identification process, media fingerprinting as well as token-based identification processes, are limited to either system identification or one-to-one identification. Current one-to-one identification mechanisms require huge volume of templates of known users, and cannot be applied to novel users. This study proposed a psychosocial approach that integrates the composition of human Polyphasia tendency into online identification processes for a one-to-many identification process. To achieve this, the study administered a Polychronic-Monochronic tendency scale measurement instrument to staff members of a research unit in a university, and the server-side network traffic of each respondent was monitored and collected in eight-months duration. A logistic model tree—after an initial classifier exploration process—was adapted for the one-to-many classification model based on human intrinsic features extracted from the network traffic and Polyphasia dichotomy. High degree of reliable accuracy of > 80% was achieved which suggests a reliable model that supports the underlying hypothesis of the proposed model. Based on this accuracy, the approach finds practical relevance in online profiling process for online identification as well as online demographic profiling for e-commerce and e-learning. Furthermore, this approach can be applied to improve recommender systems in areas such as prediction and profile delivery through the extraction of the purpose of online surfing.