Research On Approximated Sequnential Patterns Mining Algorithm Based On Bitmap

Most proposed sequential pattern mining algorisms are based on exact matching.However, data stream environment mining need more efficient algorism and make manytraditional algorism disable. We need more efficient and effective algorism to minesequential pattern in data flow. This paper focus on mining approximate sequential patternwhich can be applied to web accessing analysis, DNA sequential pattern analysis and soon.First, we proposed a bitmap based sequential pattern mining algorism called BCAF. Itintroduced sequence cluster and multiple sequence alignment to sequential pattern mining.And all operations in BCAF are based on binary bitmap. It is better than other algorismand good split between mining efficient and mining quality.Second, we also proposed a bitmap based approximate time-interval sequentialpattern mining algorism called BMATS in this paper. It applies sequence cluster andmultiple sequence alignment to time-interval sequential pattern mining once again. TIATis a new data structure we proposed to make multiple time-interval sequence alignment.BMATS is a efficient and effective algorism.At last, we programmed the two algorisms mentioned above. And detail analysis ofspace and time efficient are given. Further more, we discussed the output quality of thetwo algorisms.Experiments demonstrate that the two algorisms this paper proposed are efficient andeffective in data stream environment. Their performance outperforms obviously troditonalalgorism. And we achieve the expectant goal.