Generating Dynamic Higher-Order Markov Models in Web Usage Mining

@incollection{citeulike:3032510, abstract = {Markov models have been widely used for modelling users’ web navigation behaviour. In previous work we have presented a dynamic clustering-based Markov model that accurately represents second-order transition probabilities given by a collection of navigation sessions. Herein, we propose a generalisation of the method that takes into account higher-order conditional probabilities. The method makes use of the state cloning concept together with a clustering technique to separate the navigation paths that reveal differences in the conditional probabilities. We report on experiments conducted with three real world data sets. The results show that some pages require a long history to understand the users choice of link, while others require only a short history. We also show that the number of additional states induced by the method can be controlled through a probability threshold parameter.}, author = {Borges, Jos\'{e} and Levene, Mark }, citeulike-article-id = {3032510}, comment = {Method that generates a 2nd order Markov model adaptively starting from a 1st order Markov model. Used for modeling browsing patterns in websites. In practice the 2nd order model has only 2x or 3x the number of states of the original model, depending on a parameter that controls how much approximation error to accept from the 1st-order Markovian assumption when evaluating paths of 3 nodes.}, doi = {10.1007/11564126\_9}, journal = {Knowledge Discovery in Databases: PKDD 2005}, keywords = {clicks}, pages = {34--45}, posted-at = {2008-07-22 11:52:07}, priority = {4}, title = {Generating Dynamic Higher-Order Markov Models in Web Usage Mining}, url = {http://dx.doi.org/10.1007/11564126\_9}, year = {2005} }

TY - CHAP ID - citeulike:3032510 L3 - citeulike-article-id:3032510 TI - Generating Dynamic Higher-Order Markov Models in Web Usage Mining JF - Knowledge Discovery in Databases: PKDD 2005 SP - 34 EP - 45 N2 - Markov models have been widely used for modelling users’ web navigation behaviour. In previous work we have presented a dynamic clustering-based Markov model that accurately represents second-order transition probabilities given by a collection of navigation sessions. Herein, we propose a generalisation of the method that takes into account higher-order conditional probabilities. The method makes use of the state cloning concept together with a clustering technique to separate the navigation paths that reveal differences in the conditional probabilities. We report on experiments conducted with three real world data sets. The results show that some pages require a long history to understand the users choice of link, while others require only a short history. We also show that the number of additional states induced by the method can be controlled through a probability threshold parameter. KW - clicks N1 - Method that generates a 2nd order Markov model adaptively starting from a 1st order Markov model. Used for modeling browsing patterns in websites. In practice the 2nd order model has only 2x or 3x the number of states of the original model, depending on a parameter that controls how much approximation error to accept from the 1st-order Markovian assumption when evaluating paths of 3 nodes. AU - Borges, José AU - Levene, Mark PY - 2005/// UR - http://dx.doi.org/10.1007/11564126_9 ER -