Research On Application Of Text Mining And Recommendation Algorithm In Agricultural Information Service Platform

Science and technology have played a key role in increasing agricultural incomes and improving farmers' technological literacy.With the rapid development of information technology,a large amount of agricultural information resources have accumulated on the Internet.However,due to the lack of knowledge and application skills of farmers on the Internet,the utilization rate of agricultural resources is not high,and the promotion rate is low.For the purpose of effectively using the existing agricultural information resources to adapt to the actual reading needs of agricultural users and solve the problems of insufficient accuracy in extracting features based on content recommendation algorithms and traditional recommendation algorithms for agricultural products,this paper proceeds from the following three aspects.First,document similarity calculation method based on keywords is improved.The traditional keyword-based document similarity calculation method is research and analyzed in order to improve the accuracy of the keyword-based document similarity calculation method,and based on the original method,by introducing a variety of keyword extraction methods and the average ways are taken to improve the accuracy of document similarity.Finally,experiments verify the accuracy of the improved algorithm.Second,special stop word list of agricultural technology is established.Due to the unique professional fields and writing habits,agricultural technology documents will generate some unique stop words such as “field,breed” and similar words.These words may appear frequently but have no practical opinions on agricultural technical documents,and due to the fact that the published stop word vocabulary has not been specifically targeted at the agricultural sector,a professional stoppage vocabulary needs to be constructed for the agricultural sector.Third,a multi-featured agricultural web document recommendation method is studied.This paper designs and implements a multi-feature-based agricultural web document recommendation method.This recommendation method calculates the similarity of documents from the three dimensions of solar terms,topic characteristics,and keyword characteristics,and recommends a document set with the highest degree of similarity.The user,the document evaluation method is compared with the content-based recommendation method and the content filtering-based agricultural information recommendation method.Experimental results show that the multifeatured agricultural text document recommendation method proposed in this paper has a high recommendation accuracy rate.In summary,this paper presents a combination of browsing,collection records,keywords and multi-featured agricultural web methods about 24 solar term.The algorithm calculate the similarity from three dimensions: the user's past browsing,collection of articles and the theme feature of current reading document,the feature of keywords and solar term.Then it calculates the comprehensive similarity of documents by adding the calculation results,recommending the most similar results to users.The experimental results show that the algorithm about the recommendations of agricultural web documents is more accurate,and it has reference value for the recommendation of agricultural products.