Probability Model-based Service Recommendation In Pervasive Environment

With the development of science and technology, more and more devices, withcomputing ability, are developing with the direction of miniaturization and embedded,computer-centric computing model and desktop model cannot overcome the trouble of oneperson with more than one machine, so these face severe challenges. Pervasive computingenvironment gets widespread attention because the feature of "transparent" and "anywhere".In pervasive computing environment, the "transparent" feature is not completely refers to thephysical visibility, more important is whether the interaction between the user and thecomputer is perceived. A necessary condition to achieve this "transparent" interact is usingcontext information in ubiquitous computing environment. As an application,Web servicesare able to meet the user’s needs, but also get the attention of the researchers and the needs ofthe user, it has a self-contained, self-describing, modular feature,which makes users morehope that the Web service to be available everywhere. Therefore, to provide personalizedservice in pervasive computing environment become a research hotspot.For the application requirements of the service recommended in the pervasive computingenvironment, after analysising the technology of traditional service recommendation andservices recommended technology which integrates context, this paper mainly complete twoaspects of work as following:(1) Research context information and Web services information in ubiquitous computingenvironment, in the different areas of Web services, different context has different influenceon the services recommendation. Web services can be described in different ways, theinteraction between service requestor and service provider was realized by a series ofpublishing and binding protocol. Then develop applications for the smart mobile device toobtain contextual information, and extract key part of Web service description informations,to find service and call faster.(2) With analysising the relationship between context and services as well as thecharacteristics and application of the hidden Markov model, the paper designs a servicerecommendation mechanism based on the double hidden Markov model (Double HiddenMarkov Model, DHMM). By obtaining and processing context data, calculate the parametersof DHMM, so as to achieve the personalized service recommendation. Meanwhile, with thedynamic changes of the context information, the model parameters are updated in real time,so as to improve the accuracy of services recommendation. The simulation is based on thecharacteristics of mobile users in campus environment, by sensing user’s location, time and professional context information, and calculating the probability of service invocation fromthe service calling history, then verify the accuracy rate by calculating the recommendeddegree of personalized service recommendation mechanism based on hidden Markov model.The symbolic of the context information used in the simulation model, enhance the scalabilityof the model.