Article Preview
Top1. Introduction
Context-awareness has evolved and applied as a key component of ubiquitous and pervasive computing systems until the early 1990s. Throughout the past decade, the attention on context-aware computing has developed from desktop, web applications, mobile computing, and Internet of Things (IoT), of ubiquitous/pervasive computing. The term context-aware was first used by Schilit and Theimer in 1994, and it defines the location, object, and identity of the people who are nearby. Dey and Abowd (2003) gave the most commonly used definition in 2000. The context represented as “any information that can use to characterize the situation of an entity. An entity is a person, place, or object that is considered relevant to the interaction between a user and an application, including the user and applications themselves” (Dey & Abowd, 2000). From the smart objects (sensors), the data will acquire directly, which is inaccurate for the provision of customized intelligent services even though it includes noise and mistakes.
Consequently, contexts acquired from such objects will be biased, and may the result is inaccurate, to overcome this issue, such raw context data can be handled in favor of acquiring important data of high-level context (Hussein, 2015). Also, context recognition makes machine-to-machine communication easier as it is the main element in IoT perception.
Context reasoning is one of the most challenging functions of current research work is the way of using context information effectively. A more accurate context-reasoning definition is the deduction from the different sources of context-data of new and applicable information for the user of the users and applications. Though, here we need to highlight the context that is raw sensor data (low-level context) change into high-level (mixtures of low-level) context data. In future applications, one of the primary problems is to offer smart and ubiquitous / context-aware applications that take into consideration the context of the user. The main features of the context model are the capacity to assist the operation of reasoning concerning the different context types and their characteristics. For example, Context of low and high-level is indicated in Figure 1. and this illustrates the temperature and signal can be grouped and shows the location such as Warm and At-Home.
Figure 1. Context of low and high-level
From IoT objects, it offers a conceptual context-aware structure capable of gathering, analyzing, and inferring high-level context data. The high-level context is a difficult contextual data that is obtained by mixing several nuclear machine intelligence or context with reasoning skills. Over the past century, several contextual reasoning methods have been created, extending from straightforward early models to complicated present context models. According to the features of context information, the process of reasoning can be split into two types first is specific reasoning and uncertain reasoning method (Gao et al., 2013). Context reasoning specific examples are ontology-based, logic-based, and cased-based reasoning. The Reasoning method contains some key algorithms such as TREAT, RETE, LEAPS (Lazy Evaluation Algorithm), and LFA (Linear Forward Chaining Algorithm). Between them to create an effective and scalable context-aware reasoning is the RETE algorithm, which is the best popular method. RETE is the algorithm for the development of a rule-based system, and Dr. Charles L. Forgy proposed the forward-chain pattern matching algorithm. Rete is a directed acyclic graph representing a higher-level set of rules. Rules are usually depicted by a network that matches rules (patterns) to facts (relational information tuples) at run-time. The facts hold the working memory; the inference engine for each fact generates the working memory elements (WMEs).