📄 http:^^www.cs.wisc.edu^~devise^devise^rtree.html
字号:
Date: Fri, 13 Dec 1996 17:08:07 GMTServer: NCSA/1.5Content-type: text/htmlLast-modified: Thu, 14 Nov 1996 16:40:37 GMTContent-length: 2730<HTML><HEAD><TITLE>DEVise Application Examples</TITLE></HEAD><BODY><H1>R-Tree Validation Example</H1><!-- Session is /p/devise/session/demo/rtree/orange3.tk --><!WA0><!WA0><!WA0><IMG src="http://www.cs.wisc.edu/~devise/devise/rtree2.gif"><H2>Input Data</H2>The input data:<UL><LI>consists of three separate data sets, one for each level of the tree.<LI>was produced by dumping the R-Tree after it was bulk loaded (the datain the tree is Tiger geographic data from Orange County, California).<LI>contains the attributes: page ID, level, x, y, width, height, and color.(Color is used to make it easier to distinguish the various nodes atlevels 2 and 3.)</UL><P>The level 1 input file looks like:<TT><BR>1234 1 484402.500000 763222.500000 968413.000000 284401.000000 35<BR>1234 1 1671075.500000 682544.000000 789569.000000 484602.000000 35<BR>...</TT><h2>Problem</H2>We want to determine whether the R-Tree bulk loading algorithm has produceda "good" R-Tree (not just one that is correct, but one that iswell-structured and will result in efficient searches of the data).<H2>Visualization</H2>Click <!WA1><!WA1><!WA1><A href="http://www.cs.wisc.edu/~devise/devise/model.html">here</A> to see our model of creatingvisualizations.<P>The visualization is created as follows:<OL><LI>Define the 'RTree Global Picture' window. This window simply displaysthe level 1 data.<LI>Define the 'RTree Levels 1-2' window, consisting of two piled views,one of the level 1 data and one of the level 2 data. One of these viewsis the source for the cursor (the small yellow box) displayed in the'RTree Global Picture' window. This means that the area displayed inthe entire 'RTree Levels 1-2' window corresponds to the area withinthe cursor in the 'RTree Global Picture' window.<LI>Define the 'RTree Levels 2-3' window. This again consists of twopiled views, this time for levels 2 and 3 of the tree (level 3 is theactual objects inserted into the tree). The views in this window arelinked to the 'RTree Levels 1-2' window, so that the region displayedin the two windows is the same.</OL><H2>Observations</H2>This visualization was used to find some subtle bugs in the R-Treebulk loading code. These bugs would have otherwise been difficultto find, because they did not produce incorrect results; rather, theyrepresented a suboptimal R-Tree structure that would cause inefficientsearches of the data.<P>The data as it is shown represent the R-Tree after the bugs were fixed.When the bugs still existed, it was possible to find "bad" areas of theR-Tree by examining the visualization.<P>For more information on R-Trees, see the paper <!WA2><!WA2><!WA2><A href="http://www.cs.wisc.edu/Dienst/UI/2.0/Describe/ncstrl.uwmadison%2fCS-TR-96-1301">Using Constraints to Query R*-Trees.</A><BR></BODY></HTML>⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -