Ranking Properties of Spatiotemporal RDF Data

Abstract

Based on the sorting algorithm, the authors discuss the ranking of spatiotemporal RDF data properties and attempt to improve the query efficiency of large RDF datasets. The chapter introduces three sorting algorithms in machine learning: LR (logistic regression) algorithm, GBDT (gradient boosting decision tree) and FM (factorization machines) model algorithm. After the data sorting system is completed, the authors use A/B test method to test the system. It is self-evident that the recommendation algorithm based on FM ranking is more efficient than linear regression ranking. Using the model performance evaluation index-AUC and target detection evaluation index-MAP, the effect of FM model algorithm is significantly better than the other two algorithms. Finally, based on the Hadoop open-source big data framework, the scalability and high performance of the ranking recommendation system are guaranteed. The result of the research shows the page hits increased by 98.0% in a week.