Research On Agricultural Information Recommendation Based On Collaborative Filtering And Content Recommendation Hybrid Algorithm

Different farmers need different agricultural information.Therefore,the system needs to be different from person to person.Different recommendations are given to different users.At the same time,the text information read by different users needs to be managed collectively to form a text library.According to this interest model,the recommendation text information is recommended,and there are certain defects in the recommendation diversity.Therefore,most of the information based on content recommendation is similar to the information text that users have read before.Although the text information recommended by the collaborative algorithm is diverse,the recommendations given for different users are unsatisfactory,and there are great deficiencies in personalization.At the same time,there are other problems in the collaborative filtering algorithm,such as cold startup problems,and so on.Because the algorithm mechanism of collaborative filtering is recommended based on the behavior of similar users.Therefore,the information query is only possible to calculate the similarity between the target user and after it is read by enough users.Therefore,there is a cold start problem for newly released information.The mixed recommendation algorithm can take into account the diversity and personalization of the recommendation results,but the aspect of the diversity requirement is still a collaborative filtering mechanism,so the problem of cold start of the project still exists.Considering the problems mentioned above,this thesis studies the collaborative filtering algorithm and the content recommendation method,and finds that the mixed collaborative filtering algorithm and the content recommendation algorithm two modes create a new text information recommendation method,which not only solves the problem of personalization,but also solves the problem of diversity,and also avoids the cold start problem.The data set selected in this article is from the reading of the agricultural consulting website.The evaluation indicators are Precision,Recall,and University.The experimental methods are compared with the three recommended methods as reference points.The three kinds of recommendations for comparison are: content recommendation,partition mixed recommendation,and collaborative filtering recommendation.From the experiment,it can be clearly seen that the recommended results of the mixed method in this experiment show that the content-based method is 28.1 % less than that of Prevision,Recall,and University,28.7 %,and 30.3 %,respectively;At the same time,the comparison in the collaborative filtering method was 7.6 % higher than the Prevision,Recall method.Although there are no obvious advantages in the comparison of mixed recommendations with precision,Recall,and University,the method described in our paper does not need to wait for candidate agricultural information to accumulate enough user clicks when recommending.The problem of cold start in the mixed recommendation method is a method with good application prospects.The experimental results show that the proposed method can meet the requirement of user reading in terms of diversity and personalization,and can avoid the recommended cold start problem effectively.