This thesis is devoted to solve problems faced in developing network applications in cyber-physical environments using multi-attribute based data modeling and methodology. Recent advancements in embedded systems, sensors and wireless communication technologies have led to the development of Cyber-Physical Systems (CPSs). As an integral component of CPSs, Wireless Sensor Networks (WSNs) offer great potential for developing the new network applications. However, due to the complicacy of the cyber-physical environments, there exist many influential factors/attributes that need to be considered in these new applications. Comprehensive consideration and utilization of multiple influential factors/attributes to make better decisions or seek the optimal solution has become an essential problem. To address this issue, we first summarize the multi-factor/multi-attribute problems into two types: the multi-attribute ranking problem and the multi-attribute optimization problem. And then, we focus on these two types of problems in networked cyber-physical environments, and study the multi-attribute decisionmaking (MADM) and learning approaches in three network applications: the traffic information collection and vehicle navigation in Intelligent Transportation Systems (ITSs), and data clustering in the large, dynamic distributed networks. Firstly, the traffic-monitoring system is studied. A flexible urban traffic information collection framework based on WSNs is presented and verified. A novel usercustomizable data-centric routing scheme is proposed for traffic information delivery, in which multiple routing-related information is considered for decision-making to iii meet different user requirements. Simulations show the good performance of the proposed routing scheme compared with other traditional ones on real-world urban traffic networks. Secondly, the vehicle navigation system is studied in urban traffic networks. A novel WSN-based real-time vehicle navigation algorithm is proposed. The hybrid MADM method is presented for the real-time navigation decision-making. A new general distance metric is defined for the processing of both exact and fuzzy data. Simulations show the suitability and efficiency of the proposed algorithm. Finally, attribute-weighted distributed data clustering is studied and a novel collaborative clustering algorithm based on distributed Peer-to-Peer (P2P) networks is proposed. The attribute-weight-entropy regularization technique is used to obtain good clustering results and yield the optimal attribute weights. The kernel method is utilized in the proposed clustering algorithm to meet the needs of ‘non-spherical’ shaped data clustering. Experiments on synthetic and real-world datasets demonstrate the efficiency and superiority of the proposed algorithm.