The Study On The Recommending Methods For Online Travel Websites Association Rules

Internet data mining association rules is extremely popular research field at present. Due to the usage of association rule mining algorithm can detect potential association between the goods and then recommended to the user, it can help consumers to select good results, in increased sales and delight the user experience. This paper has studies the association rule algorithm in terms of recommendation, by presenting a recommended model based on building a block-based thinking. It, applied it to the online travel websites, trying to attract the guests to buy high-quality group-buying on online travel site by the recommended information. In this paper, the main contents and results are as followings:(1) An overview introduces the data mining and the related technologies to the association rule mining algorithm. This paper makes a comparison of the Apriori algorithm and FP-tree algorithm advantages and disadvantages in terms of mining. It discoveries FP-tree algorithm although the algorithm is better than Apriori algorithm efficiency, but the tree structure is extremely occupy memory. Even it can not perform complete when it faced to massive data could lead to algorithms(2) In order to overcome the shortcoming that the long waiting time for the classic Apriori algorithm mining, this paper introduces the technical foundation Apriori algorithm based on parallel processing, giving the parallel MapReduce-based Apriori algorithm. The Apriori algorithm parallelization processed by way of division, not only can ensures the consistency of its final mining results, but also can save time on overhead and space.(3) Based on the recommended-data applications online for travel websites in nowadays, this paper presents a model for the association rules way. With the actual situation, the model used the district labels as the split, divided the huge transaction data successfully into a plurality of data blocks. It made the parallel algorithm based on MapReduce of Apriori. The mining association rules generated were stored in the database. When a trigger event recommendation, it can take directly from the association rules library. This method had saved the excavation costs produced by every server searching, and at the mean while it also could enhance the user experience.(4) A simulation experiment is carried out based on Hadoop. By mining the training data set, the result shows that the improved algorithm has lower space complex and changes linearly with the growth of data. And in the distributed computing environment, with the increase of processing nodes, mining efficiency is significantly improved, which proves the superiority and extensibility of the improved algorithm.