Analysis And Research Of Rice Domain Knowledge Representation Based On Semantic Web In Cloud Environment

With fast data acquisition and rapid increase of data volume, current information resources of agriculture in Chinaare are presently weak in structuring, data utilization, sharing and reusing. Aiming to solve the problems above, this thesis tries to use semantic net and distributed computing framework based on Hadoop to replace the traditional uniprocessing mode. Taking domain knowledge of rice as case analysis, this thesis prefers to provide some research basis for sharing and reusing of agricultural knowledge and future semantic net on agricultural information.This thesis integrates Hadoop with jena semantic net. A semantic net framework is designed to run on the distributed platform of Hadoop. Running on expandable distributed system and supported by main query language of SparQL, the framework performs outstanding in extension, configuration, big data oriented, etc. Some innovation and improvements of the framework are shown as follows:1. In the combination of cloud computing and semantic net module, part of the code are rewritten, and TripleWritable is added to metadata types to make RDF triple serializable and make framework integration possible.2. With features based on feature analysis of distributed computing framework based on Hadoop, the integrated algorithm is designed. This new algorithm integrates numerous small files of RDF to improve efficiency of cluster processing.3. Rewrite file scheduling strategy in HDFS and modify the part of fragmentation and redundancy, problem of one Triple divided into different nodes is solved to ensure accuracy.Finally, this thesis takes rice pests as an example to test and verify the distributed framework of semantic net. The framework realizes inquiry function based on semantic, simple reasoning, high processing speed with increasing data volume, etc.