Research On Service Generation And Recommendation Of Internet Of Things Based On Behavior Characteristic

With the development of Internet of Things(IoT),more smart devices and services enter into people's lives.They provide a convenient life style while changing our lives.As important research contents,IoT service generation and recommendation are becoming more and more important.People need personalized services that are consistent with their lifestyle,and they need to find the personalized services in time.Therefore,the paper focuses on the IoT service recommendation in the smart home.The research contents include four aspects,which are user habit feature extraction,service rule generation based on user's habit,identity mapping between service and sensing device,IoT service recommendation.The main innovations include:(1)The habit features extraction algorithm is proposed.We discover that the habits which users use electrical appliance could be classified into the fix-length habit and the timing habit for the first time.In addition,there are different relationships between habits,such as select,sequence,cross,parallel,inclusion,begin,finish.The habit and complex-habit extraction methods based on the definitions of activity probability are proposed.Moreover,in order to adapt to habit variations,we propose a habit change factor,and the factor self-leaining algorithm.The experiment results show that the algorithms could set the habit strength and time zone appropriately,adapt to the habit changes and extract the habit exactly.(2)The SWRL description scheme of habit is proposed.We construct the smart home ontology model,and propose the SWRL description scheme of habit which satisfys ECA model,moreover,2 SWRL description rules for simple habits and 23 SWRL description rules for complex habits are defined.We propose the framework for service rule generation and reasoning which based on habit feature.In the framework,the user,the electronic equipment,time,environment are descripted by OWL ontology,and reasoned by SWRL.Finally,we build the rule engine based on the habit feature,and proved the correct of the theories by the exection time of rules,memory utilization and confidence.(3)The IoT mapping model of multi-identifers is proposed.In order to improve the ability for addressing between services and sensing devices,we propose the mapping model,and the mapping model follows the main line of information processing.In addition,it focus on the mapping manner of identifiers in data acquisition,transmit,process,present.The mapping model includes the sensory information identifier/IPv6,metadata/resource identifier,resource identifier/resource class identifier.We build the smart home and verfy the mapping model by the delay and the packet analysis.(4)The IoT service recommendation schemes are proposed.We proposed the service recommendation scheme based on the attribute similarity for the IoT service which is based on user's habit feature.The scheme provides the service recommendation for the users who have the similar habit feature by cosine similarity of users' attribute.Moreover,we propose the service recommendation scheme based on tripartite graph with mass diffusion and use the habit feature as a dynamic tag.Based on the balance of mass,we use the positive and negative mass diffusion results on the tripartite as the recommenation result.We propose the service recommendation scheme based on attribute correlation for the cold start problem in the service recommendation.We finally verify the correctness of the theories by different kinds of experiments,and get good results.