Research Of Assistance Context-aware And The Application On Older Social Service In Crowd Sensing

With the rapid development of mobile Internet and the Internet of things, the city is in the network coverage area. At the same time, the development of sensor technology and the popularity of smart devices makes smart phones in the market are equipped with GPS sensor. Human handled devices are changing the interaction between people and the network, so that many users are involed in the situation of location-based service. Thus context-aware is gradually step into swarm intelligence from indivudal perception, accompanied with a large number of time-position datas. These datas contain a huge pratical precious. By obtaining user’s position informations and identify the activity pattern from those informations, the behavior and intention of the user can be predicted. In the mean while, cases of the elderly getting lost tends to be relatively more with the trend of aging of world’s population. It has a great significance to identify the effective pattern of user from those huge datas and apply it to older sociation service.Based on the existing theroretical research, this paper proposes a new method to overcome the problem that the detected stay points will include a certain redundancy information from the time-space datas. For user’s activity pattern mining, researchs currently are only focus on the activity position and the regular pattern of the user, while ignoring the behavior of the user. Both of the position, regular pattern and behavior are all concerned in this paper. For the discoverying od user’s periodic behavior, researchs are not seemed to provide prediction. While in this paper, not only user’s activity patterns, but also the relevant decision suopport are put forward. To solve the existing problem of indivudual stay point detecting method, a new algorithem based on the partial positive detection is proposed. In this new method a devide and combine process is adopted to detect the stay points seperately in different ways. In the view of the traditional activity pattern recognition that key points are considerated while behaviors related to points are ignored, we proposed a refined evaluation of the user behavior to identify the moving behavior related to positions. For user’s activity pattern mining, we build the largest pattern sequences through the user’s activity sequences, and works are done based on them: identify activity patterns,discovery periodicity regular, build the activity pattern knowledge graph, construct the decision support to provide real-time decision making of user behavior. Finally, systems based on mobile terminal devices are designed and implements, and applied to the older social services.Experiments show that the improved method based on the presence of partial positive detection method can effectively solve the problem of redundant information of stay points, and the accuracy and the purity of the detection results are better than the traditional stay position method. User’s behavior can be responsed better through the comprehensive consideration of the moving tarcks related to the user’s positions. Detail assessment can be provided to predict the user’s moving behavior. It proved to with a very good applicablity to applying in the older social service. And the designed systems proved to meeting the needs of the daily data collection of older and the caring for the elderly.