Research On Recommendations Of Minority Cultural Resources Service Based On MapReduce

With the development of big data and artificial intelligence,new requirements have been put forward for the protection and inheritance of minority cultural resources.In the process of protecting and inheriting the cultural resources of ethnic minorities,many excellent traditional cultural resources of ethnic minorities are facing the crisis of loss.In order to better protect and inherit minority culture and realize the sharing and dissemination of minority culture,the application of big data and artificial intelligence is the primary technical measures.Through constructing ethnic cultural resources recommended service platform,combined with the user demand for ethnic minority cultural knowledge,personalized recommend user ethnic cultural resources,can better make users more accurate and efficient access to interested in their ethnic and cultural knowledge,in order to better promote the protection of ethnic culture and heritage.Therefore,this paper studies the recommendation of ethnic minority cultural resources under the framework of MapReduce distributed computing,which mainly includes the following aspects:(1)Based on the Alternating Least Squares(ALS)on the basis of collaborative filtering algorithm.The loss of project attribute information of stealth factor is reduced by fusing item similarity on the loss function.At the same time,the cold start strategy is introduced in the model to improve the original algorithm.The improved algorithm is implemented by MapReduce computing framework on Hadoop platform.Experimental results show that compared with the traditional recommendation algorithm,the modified ALS collaborative filtering recommendation algorithm can effectively alleviate the data sparsity problem.The recommendation accuracy and calculation efficiency have been improved to some extent.(2)The Latent Dirichlet Allocation(LDA)thematic model was carried out to carry out the extraction and tagging of cultural text resources of ethnic minorities.Based on the topic mining of ethnic minority cultural resources,the user behavior log is analyzed to establish the user connection from ethnic minority cultural resources to labels.A user-resource-tag personalized user tag model is proposed.(3)According to the user label model built in(2).First,Canopy-Kmeans clustering algorithm was used to cluster users.Then the recommendation algorithm is improved by improving the user similarity calculation method on the basis of user clustering.Finally,the user clustering recommendation algorithm that improves similarity calculation is implemented on Hadoop distributed platform.As the minority cultural resource tags and user tag construction and user clustering in the algorithm can be completed offline,online recommendation can be made more quickly.Experimental results show that the proposed algorithm is effective in the recommendation of ethnic cultural resources.(4)The implementation of prototype system is recommended.Firstly,a few cultural resources recommendation prototype system was built by Hadoop distributed platform.Secondly,functional modules such as crawling,tagging,acquisition of user logs and preference analysis are designed in the system.Finally,the personalized recommendation of ethnic minority cultural resources to users is achieved.