【摘要】：針對(duì)普通客戶端瀏覽和分析大數(shù)據(jù)困難的問題,結(jié)合Spark和LOD技術(shù),以熱圖為例提出一種面向大數(shù)據(jù)可視化技術(shù)框架.首先利用Spark平臺(tái)分層并以瓦片為單位并行計(jì)算,然后將結(jié)果分布式存儲(chǔ)在HDFS上,最后通過web服務(wù)器應(yīng)用Ajax技術(shù)結(jié)合地理信息提供各種時(shí)空分析服務(wù).文中重點(diǎn)解決了數(shù)據(jù)點(diǎn)位置和地圖之間的映射,以及由于并行計(jì)算導(dǎo)致的熱圖瓦片之間邊緣偏差這2個(gè)問題.實(shí)驗(yàn)結(jié)果表明,該方法將數(shù)據(jù)交互操作與數(shù)據(jù)繪制和計(jì)算任務(wù)分離,為瀏覽器端大數(shù)據(jù)可視化提供了一個(gè)新的思路.
[Abstract]:Aiming at the difficult problem of browsing and analyzing big data on the common client side, this paper presents a visualization technology framework for big data by combining Spark and LOD technology, taking thermal graph as an example. Firstly, the Spark platform is used for layered and tile parallel computing, then the results are distributed stored on the HDFS. Finally, the web server uses Ajax technology and geographic information to provide all kinds of spatio-temporal analysis services. In this paper, the mapping between data points and maps, and the edge deviation between tiles caused by parallel computation are mainly solved. The experimental results show that the method separates the data interactive operation from the data rendering and computing tasks, and provides a new way to visualize big data on the browser side.
【作者單位】： 浙江工業(yè)大學(xué)計(jì)算機(jī)科學(xué)與技術(shù)學(xué)院;浙江大學(xué)CAD&CG國家重點(diǎn)實(shí)驗(yàn)室;
【基金】：國家自然科學(xué)基金(61303141,61272302) 浙江省自然科學(xué)基金(LY12F02034) 國家“八六三”高技術(shù)研究發(fā)展計(jì)劃(2015AA016404)
【分類號(hào)】：TP391.41
，

本文編號(hào)：2399530