Online Product Visual Sequence Analysis

Abstract

In the upcoming several diverging fields, Sequential pattern mining automatically finds its usage. Because of this situation two challenges arise. Firstly, a huge amount of patterns are generated due to the existing algorithms which are noise in the user's view. Secondly, as datasets increases, mining all patterns become more costly. Thus, an algorithm is required which can resolve both problems and provide optimal results. This problem is resolved by including visuals along with the sequential pattern mining to create a transparent working model. We provide a neat path to visualized sequence mining allowing user to control the execution through visuals. Our path includes,1. Providing local constraints during mining process which improves users’ capacity to purify the search space without requirement of restarting. 2. Allowing stepwise visualization.3. Provides ability to control the algorithm flow. We confine our path on two event sequence dataset i.e. composing web pages visited and individual activity sequences.