Research Of The Access Of Agricultural Information Resource In Big Data Environment

With the fast development of China's agricultural,a considerable amount of agricultural data is accumulated after long term's research in agriculture sciences and analysis of relevant data.At the same time,the widely used distributed big data platform-Hadoop has been applied at so many domains that this paper proposed to take Hadoop as the data processing platform to storage and deal with the existing agricultural big data,under the consideration of features of Hadoop and characteristics of agricultural data in our country.Main works of this paper are as follows:(1)Analysis of agricultural big data.This paper analyses the current situation of the large volume of spatial data in agricultural both at home and at abroad in detail,and summarizes and analyses characteristics of agricultural data sharing in China.(2)Deign of storage model and index structure of agricultural big data.This paper designs an agriculturaldata storage structure according to types of different data based on characteristics of agricultural data in China summarized in this paper,and builds a strategy around different agriculturaldata storage structure.Then,this paper refines the traditional indexing structure to a two level indexing structure,which could improve the index between the data storage space and data partitions and the index between data partitions and data nodes which are the real data storage space in Hadoop,in order to offer a unified data storage platform for agricultural data with different types.(3)The improved data partitioning method of k-Nearest Neighbor query algorithm,and the redesign algorithm of this query algorithm.To enable the high efficient query against the large agricultural data storaged on the above distributed database,this paper refines the widely used traditional k-Nearest Neighbor query algorithm through replacing the previous regular region segmentation approach by an irregular partitioning method,and parallelized the irregular partitioning method based k-Nearest Neighbor query algorithm according the runningmechanism of the big data processing frame – Map Reduce,then the improved k NN algorithm is implishmented on Map Reduce framework to improve the overall performance.