A Big Data Query Optimization Framework for Telecom Customer Churn Analysis

One of the most popular social platforms is Twitter and telecom companies are using different algorithms and techniques for doing customer sentiment analysis over big pools of data coming through it. The biggest challenge is tweets are unstructured and queries should run accurately and fast. Our research introduces a new query analytical framework capable to tackle the big data challenges by leveraging deep learning technique, recurrent neural network and metaheuristic approach, spider monkey optimization algorithm. DeepRNN is used during tweet classification as they are capable to work with big data, and spider monkey optimization is applied for training network weights to optimize speed and get accurate results. Experimental results are compared with the existing deep convolutional network model and results show that complex analysis is conducted with good flexibility and performance. The model has optimized the training time of recurrent networks to a minimal value, i.e. 0.3 s.