173.txt

This complete matlab for neural network

发信人: yaomc (白头翁&山东大汉), 信区: DataMining
标  题: [合集]偶有些想法，大家帮着看看？
发信站: 南京大学小百合站 (Sat Dec 29 16:36:21 2001), 站内信件

zc (民工——编号0133045) 于Tue Dec 18 16:50:55 2001)
提到：

偶在想，对于传统操作数据库可不可以进行连续的挖掘呢？
例如，对于不断更新的操作数据库每隔一段时间（例如一天
或者一星期）进行一次完整的挖掘，在此基础不断的上进行
增量挖掘。可以形成挖掘出的知识库，根据这些知识库联合
操作数据库，进行连续的决策支持，并对挖掘出的知识的正
确性进行评价，对已经实施的决策进行评估。并实施对决策
的修改。
且不论软硬件代价，这个思想是否可行呢？还是早有类似的
理论？
望有识者告知！

greenflower (小呆) 于Tue Dec 18 19:15:57 2001)
提到：

i have read some paper on 增量式更新数据挖掘,
i think u'd better find some to read...



zc (民工——编号0133045) 于Tue Dec 18 20:12:12 2001)
提到：

对的，
增量挖掘是一定会使用到的。
但是，普通的增量挖掘可以针对操作数据库么？


greenflower (小呆) 于Tue Dec 18 22:13:45 2001)
提到：

the aim that this method is presented is to resolve changeable database,
i think it should operate on database... 
in details, i think u may read some on it, then u can know well...



fervvac (高远) 于Wed Dec 19 14:18:23 2001提到：

There are quite some work on incremental data mining, for example, keeping
the negative boundary information in the last session. And I see no major
difficult to apply it to the operational database, am I right?

However, if the update is constrained, better algorithms might be found. But 
I guess this idea is easy to think of and must have been done by someone, :-)

BTW, do you mean "operational DB"?



zc (民工——编号0133045) 于Thu Dec 20 20:29:44 2001)
提到：

yeah,i mean "operational DB".
but when we talk about OLAP,we say it's hard to operate on operational DB.
why can we say DM on operational DB is OK?
and i think my idea is straightforward,but hard to archieve.
what i want to know is not incremental DM to get some certain knowledge,but kn
owledge base.
and  using the knowledge base to verify knowledge and decision-making.
if you have some papers about this topic,can you send it by mail?
my mail is [e-mail].
whole paper or link are both OK. 
thank you for attention to my topic.