The Research Of Supervised Self-feedback Mechanism And Context Inconsistency Resolution

When human beings interact with other people or the environment nearby,they are good at taking advantage of contexts like gestures,expressions and situations.even without consciousness.On the contrary,for computers,such kind of ability to make use of implicit information is unavailable.This disparate situation lasted until Mark Weiser put forward the concept of pervasive computing in 1991.Since then,a new computing mode known as context aware computing has been surveyed.aiming at providing software system with awareness and realizing transparent service ultimately.Ideally the context information should be accurate,effective,non-redundant and consistent.However,in practical application,contexts descripting the status of the same object are probably provided by multiple sources from various channels.Due to the difference of precision,location and angle of sensors,as well as factors like network delay,the multi-source information is not always consistent.Before the further usage,these inconsistent contexts should be effectively processed through context inconsistency resolution.The main contents of this thesis can be summarized as four parts:1.Contexts have innate quality,which is called quality of context(QoC).This thesis focuses on the research of data sources for QoC assessment.an important but usually neglected issue.A new argument is put forward,holding the view that the inaccurate QoC evaluation should be blamed for the failure of QoC-based algorithms to obtain their theoretical performance.2.This thesis makes a new attempt to conduct quantitative research of the influence characteristics of trustworthiness evaluation on the performance of context inconsistency resolution.A new indicator referred to as decision accuracy is adopted to make more appropriate and efficient measurement to the performance of inconsistent information processing.providing a better basis for the subsequent simulation analysis.3.This thesis presents a new method of trustworthiness evaluation by utilizing the output data of the inconsistency elimination unit as the reference to assess the irnput information,which is called the self-feedback mechanism.Based on self-assessment,the data volume for trustworthiness computation is immensely increased.Thus the evaluation value tends to be more precise,and better inconsistency elimination performance is achieved at the same time.4.This thesis presents a supervisory mechanism based on voting strategy in order to ensure high reliability of self-feedback data.The performance of self-assessment can be optimized with a good tradeoff between its quantity and quality.This thesis emphasizes and proves the importance of QoC evaluation,and it puts forward the idea that context processing can be benefited from more accurate QoC assessment.The proposed trustworthiness evaluation mechanism based on supervised self-feedback creates a benign circulation between the QoC evaluation unit and the QoC application unit.Both sides benefit from this bidirectional information exchange,leading to a more precise trustworthiness assessment and better inconsistency processing as well.Furthermore,the proposed self-feedback mechanism operates in an automatic manner,which is helpful to a better user experience with less manual intervention.The above research provides an effective solution for trustworthiness evaluation in the context aware systems,and better context inconsistency resolution can be achieved at the same time.