·496· 北京科技大学学报 2006年第5期 [4]Casa-Garriga G.Discovering unbounded episodes in sequential [6]Harms S.Deogun J,Saquer J.et al.Discovering representa- data//Proceedings of the 7th European Conference on Princi tive episodal association rules from event sequences using fre ples and Practice of Knowledge Discovery in Databases.Cav- quent closed episode sets and event constraints//Proceedings of tat-Dubvrovnik,2003:83 the 2001 IEEE International Conference on Data Mining.Cal- [5]Hoppner F.Discovery of temporal patterns-leaming rules ifornia,2001:603 about the qualitative behaviour of time series//Proceedings of [7]Kosaraju S R.Fast pattern matching in trees//Proceedings of the 5th European Conference on Principles and Practice of the 30th IEEE Symposium on Foundations of Computer Sci- Knowledge Discovery in Databases.Freiburg,2001:192- ence.New York,1989:178 203 Mining algorithm of frequent episodes in an event sequence based on generalized suffix-tree QU Wenlong1.2),YANG Bingru2),ZHANG Kejun2) 1)Shijiazhuang University of Economics,Shijiazhuang 050031,China 2)Information Engineering School,University of Science and Technology Beijing,Beijing 100083.China ABSTRACT In order to mine frequent episodes from an event sequence efficiently,an algorithm based on generalized suffix-tree was proposed to discover and store frequent episodes,which uses the concept of gen- eralized suffix and contains only frequent episodes'nodes.The occurrence list of frequent episodes was used layer-upon-layer to improve the efficiency of the tree.The algorithm make full use of the order character of an event sequence and may discover the variety of frequent episodes.Experimental results show that the proposed algorithm is superior in runtime to Apriori-like frequent episodes mining algorithm. KEY WORDS event sequence;frequent episodes;data mining;generalized suffix tree. 4 9 6 . 北 京 科 技 大 学 学 报 2 0 0` 年第 s 期 [ 4 ] C sa 一 G a r iga G . 压 vosc e ri n g u n 场 nu dde e P i s《 x」se i n Se q u e n t ila d a t a / P o e d i n g s fo t h e 7 t h E u ro Pe a n Cb n f e r nce e no P h n e i - p l e : na d P r ca t i e e o f K now lde g e 肠 sc o v e斗 i n D a t a b sae . Ca v - t a t 一 D u b v ro vn ik , 20 0 3 : 8 3 H o p p n e r F . 以 cs o 理斗 o f t em 卯 r al 囚t t em -s lae m i呀 ru les a场 u t t he q u al i t at i v e be h a v olu r of t lme se it es / P ocr e d i吃 s of t h e s t h E u or eP a n C冶n fe r e n e e on P ir n e iP l e s a n d P r a e t i e e o f K n o w lde g e D i cos v e r y i n aD t a ha s e s . F r e ib u r g . 2 00 1 : 1 9 2 一 20 3 〔6 ] [ 5 ] H a rm s S , 】) 泊 g un J , aS q u e r J . e t al . D i cos ve ir 呀 re P ~ n t a · t i v e e p i汉 x l a l ~ i a t oln ru les f orm e v e n t 卿 ue n e es us i n g f r e - q u e n t e los de e p i 别 x l e se t s a n d e ve n t co n s t r a i n t s / P ocr e d i吃 5 o f t h e 2 0 0 1 IE E E I n t e rn a t io n a l oC n f e r e n e e on aD t a M i n i n g . C a l - iof m i a . 2 0 0 1 : 6 0 3 K o as r aj u 5 R . F as 、 p at t e m m a t e hi叱 i n t r 。 / P ocr e de i n g s of t h e 3 0 t h IE EE S卿 p o s i u m o n oF u n d a t i o n s o f C冶m p u t e r 段i - e n c e . N e w Y o r k , 19 8 9 : 1 7 8 M i n i n g a l g o r i t h m o f f r e q u e n t e p i s o d e s s u f fi x 一 t r e e e v e n t s e q u e n e e b a s e d o n g e n e r a li z e d Q u we , l o n g ` , 2 ) , 以N G B i n g r “ 2 ) , 刀认 N G ejK u 。 2 ) l ) S hij iaz h u a n g U n i v e sr i t y of E e o on m i e s , S hij iaz h u a n g 0 5 0 0 3 1 , C 瓦an 2 ) I n of arm r i o n E n g i n e r i n g 反h co l , U n ive rs i t y o f cS i cen e an d T ce h n o 】。 g y Be ij i n g , Be ij i n g 10 00 8 3 , C h i n a A BS T R A C T I n o r d e r t o m ine f r e q u e n t e p i os d e s f orm a n ve e n t se q u e n e e e f fi e i e n t l y , a n a lg o r i t h m b a se d o n g e n e r a li z e d s u f fi x 一 t r e e w a s p or 因 s e d t o d i s co v e r a n d s t o r e f r e q u e n t e p iso d e s , w h i e h u se s t h e co n e e p t o f g e n - e r a li z e d s u ff i x a n d e o n t a i n s o n ly f r e q u e n t e p iso d e s ’ n o d e s . T h e o e e u r r e n e e li s t o f f r e q u e n t e p iso d e s w a s u s e d l a y e r 一 u P o n 一 l a y e r t o im p r o v e t h e e f fi c i e n e y o f t h e t r e e . T h e a lg o r i t h m m a k e f u ll u s e o f t h e o r d e r e h a r a e t e r o f a n e v e n t s e q u e n c e a n d m a y d i s co v e r t h e v a r i e t y o f f r e q u e n t e p i s o d e s . E x P e r im e n t a l r e s u l t s s h o w t h a t t h e p or P o s e d a lg o r i t h m 1 5 s u P e r i o r i n r u n t im e t o A p r i o ir 一 li k e f r e q u e n t e p iso d e s m i n i n g a lg o r i t h m . K E Y W O R D S e v e n t s e q u e n e e ; f er q u e n t e P iso d e s : d a t a m i n i n g : g e n e r a li z e d s u ffi x t r e e