Design And Implementation Of Contextualized Information Delivery Architeture

Current web service technologies together with the advances in mobile devices have significantly increased the ability to provide the right information to the right person.In this trend,context-aware service has attracted lots of attention and become a hot research issue.Meanwile,an increasing number of mobile users demand adaptable services tailored to their specific requirements in a particular situation.This demand has posed a major challenge for designers and researchers investigating context-awareness:What are the solutions that develop a context-aware architecture and enable applications to adapt their behavior to mobile users' current context but without consuming too much of their attention?Our research is motivated by tackling the above challenges.We aim at offering a generic architecture supporting delivery of contextualized information that can assist mobile users to take decisions during their activities.In contrast to the traditional office work environments,the contextual situations of mobile users are characterized by change and can be described by various dimensions of context,including the physical/external context that can be captured by hardware sensors,such as location,time and speed etc.,and the logical/internal context that is related to user's activities,goal and even emotional state.In order to meet specific information needs of mobile users in a given context,context-aware applications should not only detect a change of context,but also determine the relevant information or services,which could support their task at hand.Typically,mobile police officers need technical support in the form of information systems tailored to their contextual information needs.It has important theoretical meaning and practical value to do research on context-aware services for mobile users,in order to enrich and develop the theory of dicsion support system,and also improve the ability of rapid response to deal with emergent events.The following specific requirements should be taken into account for context-aware systems that support mobile police work:(1)a routine situation of police officers can at any given time evolve into an urgent one.(2)Information needs of police officers are not only driven by their task at hand but also influenced by their targeted events and objects.(3)Police officers in particular need to be aware of information related to their personal safety.Within the collaborative projects,in the context of which the presented research was done and which is aimed at developing systems that enhance situation awareness of emergency responders,police officers involved in mobile police work are considered as our end-users.Through a questionnaire-based study aimed at these end-users on the basis of a number of realistic scenarios,we identified the specific information needs of police officers.Following the requirements derived from this study,we proposed the strategies for designing architecture for delivering relevant information contextualized to the situations of mobile users.There exists a rich body of research efforts on providing context-aware services,ranging from traffic routing to virtual tour guides.However,only a few approaches offer generic frameworks that are applicable to different domains.Moreover,despite the fact that most researchers refer to the importance of using various contexts,the primary context information used today is the user's location.Besides,we rarely see research work that considers both the location and the type of activity that a user is involved in.To our knowledge,among those context-aware systems designed by taking into account the specific requirements of emergency responders in critical situations,almost none of them address the issue of computing of the relevance of information items in a given context.In addition,they are rarely quantitatively evaluated in terms of their properties.In order to tackle the above bottlenecks,this thesis presents a generic architecture(CIDA)for contextualized information delivery relying on Semantic Web technologies,which can support mobile users in decision-making in critical situations.The main research work includes:(1)based on our questionnaire-based study,we identified and represented the information needs of police officers in the context of dealing with small-scale events.(2)Following the requirements derived from user studies,we proposed the strategies for our design of architecture for providing relevant information to mobile users in a given context.(3)We presented an ontology-based context model,which can represent the dynamic contextual situations,consisting of a generic ontology and a domain-specific ontology.(4)With the goal of showing the feasibility of CIDA,we developed the Contextualized Relevance Assessment System(CRAS).The core component of CRAS is the Relevance Assessment Rule Engine(RARE),which can re-configure the behavior of the system by executing different rules sets.(5)We focused in particular on the quantitative evaluation of the properties of our rule-based system.The behavior of CRAS is evaluated on the basis of different rule sets and datasets with regard to different configurations.The evaluations of CRAS in terms of precision and recall with respect to a gold standard demonstrated that:CRAS is able to delivery contextualized information to support mobile users in decision-making,and thus verifying the feasibility and adaptability of CIDA.The main contributions of this thesis are:(1)we offered a generic architecture(CIDA)for delivering relevant information contextualized to the situations of mobile users.CIDA was designed to have two crucial features:one is the representation of the contextual situations of mobile users relying on an ontology-based information model,which is extensible for meeting those domain-specific requirements;another one is the flexibility of adaptation,which builds on declarative rules defined in our rule language named MRRL.Those MRRL rules,specifying which type of information to deliver in which context,can be modified independently of the code executing them,thus providing a principled way to adapt systems to new domains.(2)We implemented a system that can assess the relevance of information items according to the contextual situations of end users,and developed algorithms for relevance assessment based on RDF and SPARQL.(3)We defined a quantitative evaluation methods and metrics for assessing rule-based information systems.Based on a large of experimental data,we demonstrated the feasibility of our system in the context of mobile policy work.