Parallel algorithm research of graph search and depth learning based on data mining

At present, information technology presents exponential growth characteristics, it have entered the era of large data. Data is a strategic resource as important as self-heating resources and human resources, which implied huge economic value. How to effectively organize and deal with large data will play a huge role in the socio-economic development. The graph search and depth learning algorithms play a more and more important role in the processing of large data because of their strong ability of network analysis and feature recognition and classification. In this paper, we propose a parallel optimization method based on locality principle, synchronization cost reduction and load balancing to solve the problem of width-first search. Finally, this paper combines all the methods together, and proposes a width-first search method using heuristic search. The experimental results showed that the width-first search algorithm with parallel optimization has good acceleration effect.