Situation awareness in context-aware case-based decision support

2013 IEEE 37th Annual Computer Software and Applications Conference Workshops

2013

-Context-aware computing has been attracting growing attention in recent years. Generally, there are several ways for a context-aware system to select a course of action for a particular context change. One way is for the system developers to encompass all possible context changes in the domain knowledge. Then, the system matches a context change to that in the domain knowledge and chooses the corresponding action. Other methods include system inferences and adaptive learning whereby the system executes one action and evaluates the outcome and self-adapts/selflearns based on that. However, there are situations where a system encounters unknown contexts. In such cases, instead of one action being implemented and evaluated, multiple actions could be implemented concurrently. This parallel evaluation of actions could quicken the evolution time taken to select the best action suited to unknown context compared to the iterative approach. This paper proposes a framework for context-aware systems that finds the best action for unknown context through multi-action evaluation and self-adaptation. In a case study, we show how our multi-action evaluation system can be implemented for a hypothetical hotelier who uses the nameyour-own-price mechanism to sell his perishable inventory.

Section: B Single Context -Multiple Actionsmentioning

confidence: 99%

System Evolution for Unknown Context through Multi-action Evaluation

2013 IEEE 37th Annual Computer Software and Applications Conference Workshops

2013

“…A self-adapting context with the use of context edges (a context edge is the border between two contexts) and context spaces is proposed on [6]. Other self-adapting techniques used by context-aware system include using case base reasoning to address domain specific problems and incomplete data sets [14] and trying to address the lack of domain knowledge through self-adapting whereas [4] proposes a model where both ontological and Bayesian network probabilistic reasoning are used for context reasoning and the context is modelled using ontology. Similarly, the approach described in [15] uses fuzzy sets to allow imperfection in context that is being sensed.…”

Section: A Context-aware Application Modelsmentioning

confidence: 99%

Context-Aware Framework for Performance Tuning via Multi-action Evaluation

2015 IEEE 39th Annual Computer Software and Applications Conference

2015

Abstract-Context-aware systems perform adaptive changes in several ways. One way is for the system developers to encompass all possible context changes in a context-aware application and embed them into the system. However, this may not suit situations where the system encounters unknown contexts. In such cases, system inferences and adaptive learning are used whereby the system executes one action and evaluates the outcome to self-adapts/self-learns based on that. Unfortunately, this iterative approach is time-consuming if high number of actions needs to be evaluated. By contrast, our framework for context-aware systems finds the best action for unknown context through concurrent multi-action evaluation and self-adaptation which reduces significantly the evolution time in comparison to the iterative approach. In our implementation we show how the context-aware multi-action system can be used for a context-aware evaluation for database performance tuning.

“…The model is based on Q-Learning with a feedback loop which finds the optimal action for each state by the reward it receives from the environment for actions taken in that state. Other self-adapting techniques used by context-aware system includes using case base reasoning to address domain specific problems and incomplete data sets [19] and try to address the lack of domain knowledge through self-adaption. Similarly, the approach described in [20] uses fuzzy sets to allow imperfection in context that is being sensed.…”

Section: Related Workmentioning

confidence: 99%

Context-Aware Approach for Determining the Threshold Price in Name-Your-Own-Price Channels

Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering

2016

Abstract. Key feature of a context-aware application is the ability to adapt based on the change of context. Two approaches that are widely used in this regard are the context-action pair mapping where developers match an action to execute for a particular context change and the adaptive learning where a context-aware application refines its action over time based on the preceding action's outcome. Both these approaches have limitation which makes them unsuitable in situations where a context-aware application has to deal with unknown context changes. In this paper we propose a framework where adaptation is carried out via concurrent multi-action evaluation of a dynamically created action space. This dynamic creation of the action space eliminates the need for relying on the developers to create context-action pairs and the concurrent multi-action evaluation reduces the adaptation time as opposed to the iterative approach used by adaptive learning techniques. Using our reference implementation of the framework we show how it could be used to dynamically determine the threshold price in an e-commerce system which uses the name-your-own-price (NYOP) strategy.