Quality Aware Smart Decisions Under Uncertainty - 1682 Words

IV. QUALITY-AWARE SMART DECISIONS UNDER UNCERTAINTY This section details the framework to process the QoC at each level of the context processing chain for our ubiquitous notification assistant. Additionally, it presents a probabilistic decision support model to tackle the runtime uncertainty. A brief account of our DDN model is also presented in this section. A. QoC-awareness in smart notification assistant The actions of the smart notification assistant are influenced by the uncertainty involved while processing multiple context types at runtime, in this case the locality and availability of the caregiver. As context data is extremely volatile, it requires certain QoC attributes (e.g. precision, accuracy, timeliness, trustworthiness) to†¦show more content†¦1. It depicts the significance of QoC requirements at each level of the context processing chain, from context fetching (low-level) to the context usage (high-level). The main data quality issues for our use case at the lower level can be the place and position of the sensor and the fact that the sensor may be accidentally altered resulting in wrong events. In case of a GPS sensor to infer the location of the user, the signals may be lost indoors. False events generated from these sensor’s can pose a non-negligible impact on a user’s (patient) health. Moreover, inferring context knowledge from this data will h amper the quality of the overall context data from lower level to the higher level of context processing and knowledge generation. A lack of quality at this level may harm the reasoning and actuation properties and reduce the ability of the ubiquitous notification assistant to determine the actual situation of the patient. It can have an unexpected effect on the real-time operations of the system as the derived context knowledge may be incorrect, consequently affecting the smart decisions of the system and preventing it from offering tailored interactions based on the user’s (patient) current QoC requirements. Global attributes like Timeliness must be satisfied over the entire life time of the system. Different local attributes within a context might guarantee a global attribute. Moreover, a local attribute can