Deserializing JSON data in hadoop

Abstract

MapReduce provides an efficient programming framework for processing big data in parallel in Hadoop. On the other hand, as the digitalized data becomes bigger as the advances information of technology, deserializing big JSON data into paths in advance can benefit queries on the data. Therefore, using MapReduce framework to deserialize big JSON data into JSON paths is applicable. In this paper, we propose an efficient JSON data processing mechanism based on MapReduce framework. The mechanism includes a redesign of JSONInputFormat class and the other two Map and Reduce functions.