Semantic and Scalable Processing of CityPulse Data for Real Time Event Prediction
The semantic representation of CityPulse data induces the capability of understanding the interactions between various services operated in an IOT device environment. CityPulse provides innovative smart city applications by adopting an integrated approach to the Internet of Things and the Internet of People. It establishes a complete connection of services, events and its relations for the purpose of pattern analysis and decision making. While new technologies are continuously introduced for the development of City eco-system pave the way towards the introduction of new services and the relevant collection of data pertaining to sensory devices. With the understanding of semantics of services involved and its relevant interoperability, the RDF format of data provides an easy access to extract higher- level abstractions including aggregations, mining, query processing and visualization which in turn helps the users to make better decisions on their regular day-to-day work. In this paper, we develop a predictive model which includes query engine on a distributed environment by using PigSPARQL. The query processing runs on an inbuilt MapReduce model of parallelization which in turn leads to scalability and better performance on the retrieved information compared to traditional approaches. We have explored the RDF data for three different use cases which are directly allied with Usage Behavioral Tracking, Operational Analysis and Predictive Analytics. We have employed the time series model for forecasting new values which are relevant for individual service level decision making. To represent this, a new visualization interface has been developed which easily picks relevant event patterns and routes on smart city environment. The proposed model is evaluated with the aspects of utility, reliability and also for scalability in Event processing and prediction relevant to Smart City Data.