Research On Improved Swarm Intelligence Algorithm And Its Application In Agricultural Text Classification

With the popularization of networking and the development of artificial intelligence technology,a large amount of information is published on various agricultural websites every day.In the face of the rapidly increasing amount of text,it is becoming increasingly complicated to find the required information in the huge amount of data..As the most common data mining technology,text classification can improve the efficiency of information retrieval.In recent years,more and more swarm intelligence algorithms have been applied to classification technology,and good classification results have been achieved.In the past,inspired by natural biological and physical phenomena,researchers have proposed a series of intelligent algorithms,including particle swarm optimization and gray wolf optimization algorithms.These algorithms are used to solve complex computational optimization problems,but the convergence speed and accuracy cannot be guaranteed.The classification research of agricultural text information can improve the retrieval efficiency of agricultural information,help users to dig out the value behind the text,and has very important significance for the research field.In this paper,the particle swarm optimization algorithm and the wolf swarm optimization algorithm are studied in depth,and improved methods are proposed for the characteristics and problems of the two algorithms,and these methods are combined with support vector machines to apply to text classification tasks in the agricultural field.The main research contents of this paper are as follows:(1)An improved particle swarm optimization algorithm(GEM-PSO)is proposed.On the basis of the original particle swarm algorithm,the coyote algorithm’s regeneration-elimination mechanism and the firework algorithm’s Gaussian mutation mechanism are introduced.In the GEM-PSO algorithm,Gaussian mutation mechanism is first used to increase the diversity of the population,improve the local search ability of the population,and to a certain extent avoid the particle swarm from falling into the local optimum in the optimization process.Secondly,the new birth-elimination mechanism is used to increase the global search ability of the population,which is conducive to the algorithm to find the global optimal solution.The test function optimization experiment results prove that compared with other algorithms,GEM-PSO has stronger global search ability and faster convergence speed.(2)An improved gray wolf algorithm(IGWO)is proposed.Aiming at the problem that the gray wolf algorithm is easy to fall into the local optimal value and the convergence speed is slow,the Gaussian mutation mechanism of the firework algorithm is first introduced to make the α wolf update its position every time Afterwards,adaptive mutation is carried out to ensure the population diversity of the algorithm in the iterative process,which is conducive to the algorithm to find the global optimal solution.Secondly,improve the nonlinear control parameter a to better adjust the global search and local search of the algorithm.The test function optimization experiment results prove that IGWO has higher optimization accuracy and convergence speed than other algorithms.(3)Establish IGWO-SVM and GEM-PSO-SVM classification models based on IGWO and GEM-PSO optimization algorithms,and combine the two models with PSO-SVM and GWO-SVM in UCI public data sets,existing agricultural data sets and self-made agricultural text data Do a comparative experiment on the set.The process of constructing the agricultural text corpus and preprocessing the agricultural text when self-made agricultural text data set is explained in detail,cleaning the agricultural information obtained by the crawler program,and obtaining effective agricultural text.The work of Chinese text preprocessing is realized with python language,and the feature words of agricultural text are extracted by the method of information gain.The experimental results show that the classification accuracy of IGWO-SVM and GEM-PSO-SVM classification model is higher,and the classification effect of agricultural text is better.