📄 174.txt
字号:
发信人: GzLi (笑梨), 信区: DataMining
标 题: [合集]about Opportunistic projection
发信站: 南京大学小百合站 (Thu Oct 3 22:12:27 2002), 站内信件
minerboy (miner) 于Mon Sep 16 14:50:24 2002)
提到:
谁看过!交流一下,yyj@nudt.edu.cn
helloboy (hello) 于Mon Sep 16 15:28:16 2002提到:
me.
好像是浆很多种算法凑起来。并没有谈很具体的实现。
你的高见呢?
fervvac (高远) 于Mon Sep 16 17:19:16 2002提到:
very fast on many datasets.
helloboy (hello) 于Mon Sep 16 18:45:05 2002提到:
really?
have you done such tests?
minerboy (miner) 于Tue Sep 17 13:39:21 2002)
提到:
不知道fervvac想过没有,但我总觉得结点数的理想情况是出现次数大于支持度的Item的个
数,至于每个结点的结构,是可以考虑的,我看的算法不多(apriori和fp,这片还没完,
还有一个讲格的,那个更邪门)这个想法是不是成熟,大家讨论啊
minerboy (miner) 于Tue Sep 17 13:43:26 2002)
提到:
:)it is a paper on kdd2002,
helloboy (hello) 于Tue Sep 17 15:09:15 2002提到:
I think op can be improved further.
fervvac (高远) 于Wed Sep 18 02:26:27 2002提到:
Sure, it is kindof hybrid approaches and sometimes really opportunistic, :)
as for minerboy's question, I am not sure what your ideal case means. At the
first thought, do u mean there are only nodes or any structure that
corresponds to the output?
Theoretical result sof AR mining is still missing
and makes the situation chatotic, :)
minerboy (miner) 于Wed Sep 25 20:22:02 2002)
提到:
sorry for so late to return it, 我的意思是在看了fp算法的基础上得来的,在fp中有
的项需要在树中生成若干节点,我在想是不是可以把它们并一块去而不丢失信息:)
需要你们的指教!
这样可能内存中保存的东西最小,没准生成频繁模式集时更快!!!
我做过测试,这样会丢失信息,不知道opTree如何,正在看
minerboy (miner) 于Wed Sep 25 20:24:15 2002)
提到:
另外,请大家跟我邮件联系讨论,共同进步:)yyj@nudt.edu.cn
helloboy (hello) 于Thu Sep 26 18:07:03 2002提到:
I can't quite understand your idea.
opTree, in fact, is a hybrid method, which is not implemented detailedly.
fervvac (高远) 于Sat Sep 28 14:21:01 2002提到:
Maybe minerboy can illustrate his idea by a concrete example?
⌨️ 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 + -