Collective Classification with Relational Dependency Networks

摘要: Collective classification models exploit the dependencies in a network of objects to improve predictions. For example, web pages, topic page may depend on topics hyperlinked pages. A relational model capable expressing and reasoning with such should achieve superior performance that ignore dependencies. In this paper, we present dependency networks (RDNs), extending recent work setting. RDNs are collective offers simple parameter estimation efficient structure learning. On two real-world data sets, compare ordinary probability trees show improves performance.