Big Data processing for telecom operator system

The paper deals with approach to the processing of large data for the telecom operator based on methods of intelligent analysis, fuzzy sets theory, meta-graph theory, as well as improving e-learning methods. Telecom operators generate an amount of data (defining as Big Data). This data consists of detailed call data, network data and customer data. The approach to Big Data processing proposes designing the intellectual software tools based on flexible workflow (scenario pattern) using the ontology model that presents all groups of the mathematical methods and services. Selection of the most suitable algorithms are held by means of modified e-learning methods depending on data flow in training process. The proposed approach allows to overcome the Big Data processing problems based on fuzzy knowledge base: computational complexity for e-leaning algorithms; choosing the best clustering, decision-making and inference algorithms that depends on input data characteristics. The Metagraph Theory is used for testing fuzzy rules correctness. The effectiveness of the proposed approach is considered on the example of analyzing the quality of the services provided in the node of the telecom network.