Building Graphical Models from Relational Databases for Context-Aware Querying

Context-aware plays an important role in modern society, especially to preference queries for user preferences depend on user current contexts. Popular mobile devices such as GPS, sensors and RFIDs produce context information to facilitate user right information acquisition. Due to that relational database is a powerful and sophisticated tool to manage large amount of data efficiently, context information can be taken as data items stored in the relational databases. To evaluate context-aware queries, graphical models are built from these relational data thus context-aware queries can be realized to query the constructed graphical models. To build a graphical model, user-defined methods are time-consuming and not efficient. To solve this problem, sophisticated database techniques are thoroughly exploited as well as synthesizing powerful ideas for defining probability distributions over relational domains to learn the graphical models, then query the learned models to realize context-awareness.