Research On Data Extraction And Classfication For The Bamboo Germplasm Resource

The thesis aimed to propose the research on Bamboo germplasm resources data extractionand their classificationmethods,because there have been many problems about the basic Bamboo resources data,such as the controversial classification of Bamboo species,the incomplete statistical data and the low level of information analysis and processing utilization.On the basis of constructing the Bamboo germplasm resources database,three main studies were carried out,including automatic extraction and structuralization of Bamboo morphology data,the model for the classification of Bamboo species by the data mining,and the design and implementation of Bamboo germplasm resources data mining system.Firstly,the structurization method of Bamboo species data was studied to automatically acquire the Bamboo morphology data by using regular extraction model.Secondly,Bamboo morphology database was set as the research object,while the Bamboo species classification model was constructed by support vector machine algorithm.Finally,Java language and MySQL database were used in developing the Bamboo germplasm resources data mining system,and then the reliability and feasibility was verified in theory and methodology by real examples.The individual research content in detailed was shown as below:(1)The data structuralization method of bamboo species was constructed based on the regular extraction model.In the structuralization method,the properties of Bamboo database were used as extraction template,and the extracting regulation was constructed by using the regular expression.The information extraction system was designed and achieved upon the regular extraction model.Through the study,the bamboo morphology data was automatically extracted and structurally stored from historical document and electronic resources.(2)Support vector machine algorithm was used to study the Bamboo species classification model.ReliefF arithmetic was applied as the attributes choosing strategy.SMO function was used to construct the modeling algorithm.The grid searching technique and cross validation were combined to optimize model parameters.Base on the above mixed strategy,the comprehensive optimization of Bamboo species classification model was studied.The small sample instance of Bamboo species database was used to confirm the effectiveness of classification model.This study provided a multiple parameter quantitative analysis and data-based method for the controversial classification of Bambusoideae subfamily.(3)Bamboo germplasm resources data mining system was developed.Java and MySQL languages were used to develop the Bamboo germplasm resources data mining system on the Eclipse platform.The proposed method was achieved in the system.In this system,many functions have been achieved,such as the authority management for different users,multimode acquisition for source data,the pretreatment and classification mining of Bamboo data,and maintenance and retrieval query of database.In the present study,the application and inplementation of data mining techniques was explored in the protection,storing and analysis of Bamboo germplasm resources.Through this study,the information utilization degree of Bamboo germplasm was improved.Furthermore,the data analysis,the method and technique on the processing and optimization of Bamboo data information were innovated.It was suggested that the study showed important theoretical significance and practical value for the innovation of Bambusoideae plant classification method.