Seqdbscan : a New Sequence Dbscan Algorithm for Clustering of Web Usage Data

Web is a vast area for data mining research. It is used in finding the user access patterns from web access log. User page visits are sequential in nature In this paper,I proposed a new clustering algorithm, SeqDBSCAN for clustering sequential data.. we adopted a similarity preserving function called sequence and set similarity measure SM that captures both the order of occurrence of page visits as well as the content of pages we conducted experiments comparing the results of SeqDBSCAN with other similarity measures S3M, Euclidean and Jaccards. The clusters resulting from these measures are computed using a cluster validation technique called Average levensthein distance(ALD).. Based on these results, We tested the new algorithm on dataset namely, MSNBC dataset and proved that the inter cluster similarity is high in SM when compared to the Euclidean and Jaccards distance measures and a set of experiments are conducted to investigate whether clustering performance is affected by different sequence representations, and different distance measures and other factors like number of web pages, similarity between clusters, number of user sessions , number of clusters to form.