Ӧ
COMPUTER APPLICATIONS
1999 19 9 Vol.19 No.9 1999



һʵʱ̿еھ㷨о
ѧʤ
ժҪڹҵ̿еӦʮֹ㷺һھ򷽷оҲʮҪ塣ĽһһʵʱЧԶھ㷨Աʹʵʱ̿Զھ򣬲ھĽԤ⡣
ؼʡھ򣬹̿ƣģʽ
A DATA MINING ALGORITHM IN REAL-TIME
PROCESS CONTROL SYSTEM
Yang XuebingLiu ShengjunCai Qingsheng
Department of Computer Science, University of Science and Technology of China, Anhui.Hefei 230027
AbstractComputers are widely used in the field of real-time process control, and its very necessary to pay more attention to the research in this field. In this article, well give an efficient mining algorithm in real-time data.
KeywordsData mining, Process control, Sequential pattern
1
ݿн֪ʶKDD(Knowledge Discovery in Database)Ŀǰ˹оһӦǰһУĿǰ˴֪ʶ㷨󲿷㷨ܴݿзЧ֪ʶ㷨ĿǴӸ͵Ĵݿھ֪ʶһͨԣ㷨Чʽϵͣ㹤ҵеʵʱҪ󣬺ֱʵʱ̿СβʵʱеݲɼĿܵڷCBR(Case Based Reasoning1ݼKDDģʽ(Sequential Pattern Mining)2ݣ3ݣ4ݣ5ݵ,ǵʵʱ̿еԣʵʱ̿еھ㷨DMRP(Data Mining in Real-time Processing)
ҵ̿Ǻһ򣬼жݲɼ״̬ʵʱƵãһͬʵʱɼݣѹ¶ȡѹת١ȣ˹һǺЩݼϵģ˵ʱܽᲢԤˣüԶʵʱشӲгȡ֪ʶ򣬲ûõ֪ʶԼеĹ̽йԤ⣬ѻõ֪ʶԷƫĹ̽ʵʱôյõЧ
2
2.1ʵʱ̿Ƶص
ͨݿϵͳеݣ̳ݣдݵȣȣвԵص㣬Ҫ¼棺
ظԡһһһƷģ豸תԵԣݲɼǰһʱУÿڲɼһġ
ݵ׼ȷԡеֱӲɼĲݣ׼ȷԸߣвؿȱʡ
١̿пǵɱ⣬ɼźһ㶼Ǿѡȡģ˶Թ̿ûֱӰźͨǲᱻɼġ
2.2ʽ
̵һɼݳ֮ΪһУSequence˴ڿʼڽڼʱ(t)ĲɼֵÿʱtĲɼֵӦʱ̵һ״ֵ̬St(Status-Set),St={st1,st2,,stm}ÿҲ֮Ϊһ(Case)
ݼD={Case1,Case2,,Casen}ʾһѲɼķļϣCaseiʾiӦķ1inCasei={Sti1,Sti2,,Stik}kΪһйɼ״̬Stih={st1,st2,,stm}ʾiеh1hkst1,st2,,stmΪihʱ̲ɼĸ״ֵ̬
跶S={S1,S2,,Sv},1i<jv, SiD,SjD, SiSj䣬vDлͬеS䱾DһͶӰSi={Sti1,Sti2,,Stim}
мS={S1,S2,,Sv}Si={Sti1 U {c1},Sti2 U {c2}Stip U {cp},,Stik U {ck}},1iv1pkУcpΪ{Sti1,Sti2,,Stip}Dظܴ
㷨ĿҪǴӼDвSSɹԤ⡣Ϊ˵㣬ٶйظ£
. (casenumber),ָDеУܸ
. ͷ(mincasenum),ָûһʹݼDзڴ򲻲κι
. Уsubsequence,ָдӵһԪؿʼһԪعɵӼ
. kУksubsequence,ָԪظΪkС
. ֶ֧(support)ָDظ / Dܵķ
. ֶ֧(minsupport),ָûһеֶ֧ȴڵڴΪƵС
. ζ(confidence),һǴڿʼڽһУʱ򣩣һУSt1,St2,,Sti-->Stj,,Stk,1ijkζָkеֶ֧iSt1,St2,,Stiֶ֧ȵıֵ
. ζ(minconfidence),ûֵĳζȱڴֵΪЧЧԤδơ
. ״̬ȽϺ(compare)ݾһбӦλõ״ֵ̬ǷƵĺͨ˺αȽӦǷжǷƣ˺ֵΪ-101ֱʾ<=>
3㷨
3.1ӼDɼS㷨generateS
ʼS={},D=ĿǰѴڵCasesɵļ,
datacount=0//datacountԱDеķ
for all CaseD do //ѭɨһ
if CaseSthen//˷״γ֣
CasesSxSxcΪ0, 
//˷ɺмc
S=S U {Sx}//뼯S
fi
for all SiS do//ѭSеУ
//еĳּ
for all SiSubk do
if SubkCase k then
Siеkеck1
fi
od
od
datacount=datacount+1
od
3.2ʵʱ¼Ԥ㷨RTIUPA
һڲɼΪv
LOOP startֱ豸ֹͣ
Stemp={},ȴһڿʼ,k=0
do while k<= v
//ζһвɼĸ״ֵ̬
ȴһ״̬Stk={st1,st2,,stm},
Stemp=Stemp U {Stk U ck},k=k+1,ck=1
for all SiS do
if StempSiһkthen
SiеӦеļck=ck+1,ck=ck
fi
od
if datacount>=С then
if k<v Stempֶ֧ȣsupport
Сֶ֧(minsupport) then//Ԥ
Stemp1=Stemp
for j=k+1 to v do//δ
SҳStemp1еľcj+1j+1Stemp2
if Stemp ==>Stemp2ζ(conf)ζ(minconf) then
Stemp1=Stemp2
else
break//˳˲ѭ
fi
od
j>k,Stemp1ΪδУԤ
fi
fi
od
if StempSthen
S = S U {Sx} //·S
fi
D=D U {Stemp}
datacount=datacount+1
LOOP continue
3.3DMRP㷨
if ״ then
call generateS
else
call RTIUPA
fi
4㷨
״УɨݿһΣgenerateSSǵ豸ظԣSԪӦ٣㷨ִʱҪݿDԪظÿеĳʱ临ӶԼΪnk,nΪDkΪгȡ
RTRUPA㷨ʵʱУ㷨ЧҪ3.2еforѭѭSSһСЧʺܸߣʵʱҪ
עоõȻѧ
߼飺ѧְ˶ʿоо˹ܡ֪ʶ֡
ʤְ˶ʿоо˹ܡ֪ʶ֡
ڣʿоʦо˹ܡ֪ʶ֡
ߵλйѧѧѧϵ.Ϸʣ230027
ο
1ݡʷֲ.߼˹.ѧ磬1997,12
2ݡR.Agrawal,R.Srikant. Mining Sequential Patterns.11th,Intl Conf. on Data Engineering(DE-95),Taipei, Taiwan, March 1995
3ݡR.Srikant,R.Agrawal. Mining Sequential Patterns. Generalizations and Performance Improvements. EDBT-96,Avignon,France,March 1996
4ݡŷΪ񣬲.ݿԶֹģʽ.ѧ1997;811864-870
5ݡC.Hidber. Online Association Rule Mining. Technical Report, UCB/GSD-98-1004,Department of Electrical Engineering and Computer Science, University of California at Berkeley
ո:1999-03-18
