A Semantic++ MapReduce: A Preliminary Report

Abstract

Big data processing is one of the hot scientific issues in the current social development. MapReduce is an important foundation for big data processing. In this paper, we propose a semantic++ MapReduce. This study includes four parts. (1) Semantic++ extraction and management for big data. We will do research about the automatically extracting, labeling and management methods for big data's semantic++ information. (2) SMRPL (Semantic++ MapReduce Programming Language). It is a declarative programming language which is close to the human thinking and be used to program for big data's applications. (3) Semantic++ MapReduce compilation methods. (4) Semantic++ MapReduce computing technology. It includes three parts. 1) Analysis of semantic++ index information of the data block, the description of the semantic++ index structure and semantic++ index information automatic loading method. 2) Analysis of all kinds of semantic++ operations such as semantic++ sorting, semantic++ grouping, semantic+++ merging and semantic++ query in the map and reduce phases. 3) Shuffle scheduling strategy based on semantic++ techniques. This paper's research will optimize the MapReduce and enhance its processing efficiency and ability. Our research will provide theoretical and technological accumulation for intelligent processing of big data.