Research On Agricultural Information Search Engine Classifier

With the rapid development of the Internet, the human society has entered the era of information explosion of knowledge. This brings a wave of agricultural information, and gives a convenience in agricultural information search for farm workers. Knowledge means wealth. Agricultural practitioners gather the wealth of knowledge information, from the agricultural information. However, large amount of information are not meant to be the required information query quickly and effectively, fast positioning of agricultural field refinement information and classified search is necessary and a must.This paper takes the agricultural information search engine classifier as the research object, comprehensively introduces the current status of information text classifier, the development process of classifier at home and abroad, on the base of the classification features extraction, training samples and numerous classification algorithms, first puts forward this feature extraction method which has characteristics of agricultural information text characteristics, on the base of training, establish the agricultural information text training base. To resolve the classification effect, we use a modified naive Bias classification of agricultural information, design and implement the agricultural information search engine classification system.The world does not have two leaves as like as the same. Each object has its unique property. Text information objects also have their own unique identification features for recognition and classification. In this paper, four kinds of text feature extraction of information gain, mutual information, chi square statistics and document frequency compared with the experimental realization of algorithms are discussed, are proposed to extract the text characteristics of agricultural information: the feature extraction of the text based on document frequency, using TF-IDF, vector space model and cosine to calculate correlation, on the basis of this, according to the principles of agriculture information classification and the degree of recognition, choosing the agriculture category text information, the eventual establish the agricultural information of text training base.Any kind of classification algorithm does not have the absolute superiority, has the difference. The effect is not the same of classifier with different text information. In this paper, the experimental comparison of decision tree, K-nearest neighbor, support vector machine and Naive Bias classification algorithm for agricultural information text classification, uses and modifies Naive Bias classifier, the main improvement in two aspects: the simple calculating formula of variation of Bias algorithm, two value model is transformed into a polynomial model, polynomial model formula was established, the experimental results of data comparison; in the classifier deployment, the classifier distributed deployment to multiple computers, uses the sequencing results of the Top-N algorithm, experimental results for data comparison.According to the multiple group experimental results, in the software design theory, combined with the modified Naive Bias algorithm, this paper uses the agricultural information text training base, designs and implements the agricultural information search engine classification system, gets the result data of agricultural information text classification experiments. The experimental results show, after the modification of Naive Bias classifier in classification, the accuracy and efficiency are improved to a certain extent, this is a practical and reliable agricultural information search engine classification system.To sum up, this paper is base on the text which agricultural information search engines crawl. From extraction text characteristics of classification of information, text training to classification algorithm, this paper researches on agricultural information text classifier, through the experimental comparison, puts forward the feature extraction methods of agriculture information, establishes the agricultural information text training base, improvements the Naive Bias classifier algorithm, from the deployment, deploys distributed the classifier system, finally achieve the modification of agricultural information text classifier. This paper provides the theoretical and experimental platform basis for the agricultural information text classification. At the same time, this study can also be used as a practical application.