Research On Optimization Of Tourist Attractions Recommendation Algorithm Based On Spark Cloud Computing Platform

In the new era of tourism,traditional travel methods are undergoing profound changes.Online travel has gradually emerged and has increasingly become a new hot spot in the travel market.Nowadays,in the context of big data,online travel data is increasingly overloaded,and data storage,calculation analysis and management have become one of the urgent problems in the travel industry.In addition,the overloaded travel data also brings great information distress to users and reduces the user's sense of travel service experience.In order to solve the above problems,new platforms and suitable recommendation algorithms are urgently needed to provide users with better tourism information screening and recommendation services.Aiming at the shortcomings of current tourist attraction recommendation research,this paper proposes an optimization research plan for tourist attraction recommendation algorithm based on Spark cloud computing platform.The main research work is as follows:1.Combined with Spark cloud computing platform technology.Spark is currently an excellent cloud computing platform for big data storage,calculation analysis and management in all walks of life.Spark can use Hadoop's distributed storage platform to store overloaded travel big data information.In addition,Spark's distributed computing architecture can effectively improve the efficiency of analyzing and calculating travel data.This paper uses the Spark cloud computing platform to realize the distributed storage of tourism data and the parallel computing of the tourist attraction recommendation algorithm to improve the timeliness of the algorithm recommendation.2.Crawl the real tourism data set.The python crawler code is used to crawl the data of the tourism website,in order to obtain the user's real rating and comment text of the scenic spots.In order to prove the practicability of the improved algorithm,the crawled real data set is used in the comparison experiment of recommendation algorithm.3.This paper proposes an algorithm model of LDA topic weighting based on rating and comment text information.The algorithm uses rating and comment information to provide personalized scenic spot recommendation for users.Firstly,for the collection of comment text information,the topic distribution of the collection is obtained by the LDA topic model,and then the similarity is calculated by the redefined relative entropy method.Secondly,considering the impact of rating on each comment,the weighted value obtained from the normalized user average score is used to weight the topic distribution generated by a single comment,and the similarity is calculated.Then,the two calculated similarities are combined in a certain proportion to obtain the joint similarity.Finally,the joint similarity is used to calculate the predicted score of the target object to the unrated scenic spots,and the scenic spots are sorted and recommended according to the predicted score.The experimental results show that the proposed algorithm model has high accuracy in scenic spot rating prediction.