## Model-based clustering and visualization of navigation patterns on a web site (2003)

### Cached

### Download Links

- [research.microsoft.com]
- [www.springerlink.com]
- [www.datalab.uci.edu]
- DBLP

### Other Repositories/Bibliography

Venue: | Data Mining and Knowledge Discovery |

Citations: | 53 - 0 self |

### BibTeX

@ARTICLE{Cadez03model-basedclustering,

author = {Igor Cadez and David Heckerman and Christopher Meek and Padhraic Smyth and Steven White},

title = {Model-based clustering and visualization of navigation patterns on a web site},

journal = {Data Mining and Knowledge Discovery},

year = {2003},

pages = {399--424}

}

### Years of Citing Articles

### OpenURL

### Abstract

We present a new methodology for exploring and analyzing navigation patterns on a web site. The patterns that can be analyzed consist of sequences of URL categories traversed by users. In our approach, we rst partition site users into clusters such that users with similar navigation paths through the site are placed into the same cluster. Then, for each cluster, we display these paths for users within that cluster. The clustering approach weemployis model-based (as opposed to distance-based) and partitions users according to the order in which they request web pages. In particular, we cluster users by learning a mixture of rst-order Markov models using the Expectation-Maximization algorithm. The runtime of our algorithm scales linearly with the number of clusters and with the size of the data � and our implementation easily handles hundreds of thousands of user sessions in memory. In the paper, we describe the details of our method and a visualization tool based on it called WebCANVAS. We illustrate the use of our approach on user-tra c data from msnbc.com. Keywords: Model-based clustering, sequence clustering, data visualization, Internet, web 1

### Citations

8083 | Maximum likelihood from incomplete data via the EM algorithm
- Dempster, Laird, et al.
- 1977
(Show Context)
Citation Context ...The priors are discussed in more detail in the Appendix. We learn the parameters using the EM algorithm, an iterative algorithm that finds local maxima for the MAP (and ML) parameter estimates (e.g., =-=Dempster et al., 1977-=-). The algorithm chooses starting values for the parameters, and then iterates between an Expectation or E step and a Maximization or M step until the parameters values converge to stable values (as d... |

1039 |
Bayesian Theory
- Bernardo, Smith
- 1994
(Show Context)
Citation Context ...alues for any set of parameters will depend on the coordinate system used to express the parameters. The MAP values for a multinomial distribution expressed in the natural parameter space (see, e.g., =-=Bernardo and Smith, 1994-=-) is given by φ MAP i = ni + αi ∑a j=1 n , i = 1,...,a j + α j420 CADEZ ET AL. A.3. Our EM-algorithm implementation To describe the EM algorithm for finding a local maximum of our model parameters θ,... |

524 | Hidden Markov models in computational biology - Krogh, Brown - 1994 |

479 | Bayesian classification (AutoClass): Theory and results - Cheeseman, Stutz - 1996 |

466 | Mixture Models: Inference and Applications to Clustering - McLachlan, Basford - 1988 |

311 | Model-based Gaussian and non-Gaussian clustering - Banfield, Raftery - 1993 |

278 | Using Predictive Prefetching to Improve World Wide Web Latency - Padmanabhan, Mogul - 1996 |

277 | How many clusters? Which clustering method? Answers via model-based Cluster Analysis - Fraley, Raftery - 1998 |

166 | Footprints: history-rich tools for information foraging - Wexelblat, Maes - 1999 |

156 | Discovering Web Access Patterns and Trends by Applying - Xin, Han - 1998 |

151 | From user access patterns to dynamic hypertext linking - Yan, Jacobsen, et al. - 1996 |

132 | Strong regularities in World Wide Web surfing - Huberman, Pirolli, et al. - 1998 |

116 | Link prediction and path analysis using Markov chains - Sarukkai - 2000 |

112 | M.,"Data mining of user navigation patterns - Borges, Levene - 2000 |

109 | Selective Markov models for predicting Web-page accesses,” presented at the - Deshpande, Karypis |

102 | Speculative Data Dissemination and Service to Reduce Server Load, Network Traffic, and Service Time - Bestavros - 1996 |

71 |
Expected information as expected utility
- Bernardo
- 1979
(Show Context)
Citation Context ...f the parameters obtained from the training data, and length(xi ) is the length of the sequence for user i. Note that log scores in general have interesting properties and have been used extensively (=-=Bernardo, 1979-=-). Also note that this particular log score, which uses a base-2 logarithm and a length-of-sequence normalization, corresponds to the average number of bits required by the model to encode a category ... |

66 |
The Estimation of Probabilities
- Good
- 1965
(Show Context)
Citation Context ...ferred to as maximum a posteriori or MAP estimates. When used in conjunction with vague or non-informative priors, MAP estimates are smoothed (i.e., less extreme) versions of ML estimates (see, e.g., =-=Good, 1965-=-). In the work described in this paper, we learn MAP estimates for the parameters θ using diffuse Dirichlet priors with an effective sample size of 10 −2 . (Neither the predictive nor visualization re... |

57 | Distributions of surfers’ paths through the world wide web: Empirical characterizations. World Wide Web - Pirolli, Pitkow - 1999 |

57 |
Market Segmentation: Conceptual and Methodological Foundations. 2000
- Wedel, Kamakura
(Show Context)
Citation Context ... Banfield and Raftery (1993), Cheeseman and Stutz (1995), and Fraley and Raftery (1998). In addition, there have been numerous applications of this approach in areas as diverse as consumer marketing (=-=Wedel and Kamakura, 1998-=-) and atmospheric science (Smyth et al., 1999). Nonetheless, there is relatively little work on probabilistic model-based clustering of sequences. Rabiner et al. (1989) provide an early algorithm for ... |

51 | Discovery of interesting usage patterns from web data - Cooley, Tan, et al. - 1999 |

50 | Adaptive Web Navigation for Wireless Devices - Anderson, Domingos, et al. |

49 | Ghil,Multiple Regimes in Northern Hemisphere Height Fields via Mixture Model Clustering
- Smyth, Ide, et al.
(Show Context)
Citation Context ...1995), and Fraley and Raftery (1998). In addition, there have been numerous applications of this approach in areas as diverse as consumer marketing (Wedel and Kamakura, 1998) and atmospheric science (=-=Smyth et al., 1999-=-). Nonetheless, there is relatively little work on probabilistic model-based clustering of sequences. Rabiner et al. (1989) provide an early algorithm for clustering different speech utterances using ... |

48 | Improving the Effectiveness of a Web Site with Web Usage Mining - Spiliopoulou, Pohle, et al. - 1999 |

47 | Predicting users’ requests on the www - Zukerman, Albrecht, et al. - 1999 |

46 | Clustering of Web users based on access patterns,” presented at - Fu, Sandhu, et al. - 2002 |

44 | WebSIFT: The Web Site Information Filter System - Cooley |

30 | HMM Clustering for Connected Word Recognition - Rabiner, Lee, et al. - 1989 |

28 | Probabilistic Model-Based Clustering of Multivariate and - Smyth - 1999 |

25 | Bayesian classi cation (AutoClass): Theory and results - Cheeseman, Stutz - 1995 |

25 | A GeneralizationBased Approach to Clustering ofWeb Usage Sessions - Fu, Sandhu, et al. |

23 | Model-based Gaussian and non-Gaussian clustering. Biometrics - eld, J, et al. - 1993 |

18 | Predicting a Web user’s next access based on log data - Sen, Hansen |

13 | Finite discrete markov process clustering - Ridgeway - 1997 |

12 | Probabilistic clustering using hierarchical models - Cadez, Smyth - 1999 |

11 | Strong regularities in world wide web sur ng - Huberman, Pirolli, et al. - 1998 |

11 | Visualizing crowds at a Web site - Minar, Donath - 1999 |

10 | Efficient data mining for traversal patterns - Chen, Park, et al. - 1998 |

8 | Mixed Markov and latent Markov modelling applied to brand choice behaviour - Poulsen - 1990 |

8 | Computationally efficient methods for selecting among mixtures of graphical models - SpringerThiesson, Meek, et al. - 1998 |

7 | Clustering sequences using Hidden Markov Models - Smyth - 1997 |

4 | Clustering finite discrete Markov chains - Ridgeway, Altschuler - 1998 |

2 | how manyclusters? Which clustering method? Answers via model-based cluster analysis - Fraley, Raftery - 1998 |

2 | Improving the e ectiveness of a web site with Web usage mining - Spiliopoulou, Pohle, et al. - 1999 |

2 | Computationally e cient methods for selecting among mixtures of graphical models, with discussion - Thiesson, Meek, et al. - 1999 |

1 | Visualizing crowdsataWeb site. In CHI'99 late-breaking papers - Minar, Donath - 1999 |

1 | Link prediction and path analysis using markov chains - Saruukkai - 2000 |

1 | Cadez received a BS degree in Physics from the University of Belgrade, Yugoslavia in 1995, an MS degree in Physics from the University of California, Irvine in 1997, and a PhD in Computer Science from UC Irvine in 2002. He was a recipient of a Microsoft G - Igor |

1 | is founder and manager of the Machine Learning and Applied Statistics Group at Microsoft Research. Since 1992, he has been a Senior Researcher at Microsoft, where he has created applications including data-mining tools in SQL Server and Commerce Server, t - Heckerman |