Research And Implementation Of IAD Behaviors Detecting Model Based On Emerging Patterns

With the increasing popularity of the network, social networking, shopping network, instant messaging software also will flourish. But, because patients depend on the internet, which leads to significant psychological abnormalities and causes physiological damage phenomena, It is internet addiction problem. Namely, Pathological Internet Use, for short PIU. How to early detection and treatment of addiction is a frontier problem of industrial community and academic field facing. Currently, most studies of this problem based on psychology, sociology and medical direction, while computer technology has not been involved in this issue. Therefore, in this dissertation the internet addiction problem is studied from the perspective of computer data mining, it is put forward that internet addiction pattern mining and detection model based on emerging pattern (EP), this dissertation provides a theoretical reference basis for the further t effective treatment of internet addiction.Emerging pattern is a new pattern of contrast mining, it is the item set which support significantly change from one dataset to another dataset, it can obtain the difference between objective class and negative class, we can construct a effective classifier based on emerging Pattern. For IAD pattern mining and detecting model, at first, collecting user’s simple event of net suffering, then reasoning complex events with semantic information based on production rule, at last, mining generator according to action equivalence class, because generator significantly represent characteristic of dataset and has a simple form.This thesis puts forward two kind of PIU detecting algorithm, Generator-based PIU Detecting Algorithm-GBPDA and EP-based PIU Detecting Algorithm-EPBPDA. From the perspective of generator, GBPDA algorithm selects the generator which significantly represents IAD action, gives the diagnosis by contrasting the generator of IAD dataset and testing dataset and scoring. From the perspective of EP, EPBPDA algorithm mines jumping emerging pattern and essential emerging pattern, puts forward a scoring mechanism which considers growth rate, support, JEP, eEP and which be used for detecting IAD.Experiment is carried out based on real dataset and simulation dataset, it tests efficiency of two algorithms, such as, run time, memory requirements, and effectiveness, such as, precision, false positive, false negative. Experiment result indicates that when data size is small, two algorithms have a good detection result and because EP’s ability of differentiate is stronger than generator’s, effectiveness of EPBPDA algorithm is better. But efficiency of GBP DA algorithm is better, the reason is that mining EP need more time and memory. When data size is large, comparing to GBPDA algorithm, effectiveness of EPBPDA algorithm is significantly better, simultaneously, because the increases of the number of EP is more than the increases of the number of generator, run time of EPBPDA algorithm is less than run time of EPBPDA algorithm.