Improving Mobile Web Navigation Using N-Grams Prediction Models - Distributed Artificial Intelligence, Agent Technology, and Collaborative Applications - page 325

parameters followed by the probability threshold, while depth bound has little impact. The best predic-
tion model is the 2-gram model followed by 3-gram model and 4-gram model.
These results are comparable to results from the MINPATH algorithm with N-gram model. Moreover,
the percentage of savings are comparable to these reported by Anderson et al., (2001), though a different
data set was used. This demonstrates that N-gram models work as well as other models.
conclusion and f uture work
In this chapter, we proposed to use a simple prediction model, N-gram, for improving mobile Web navi-
gation. Our approach is implemented and experimented with two real datasets. Experimental results
show that N-gram is as effective as more complex models used in other research in predicting useful
shortcuts. An interesting finding is that 2-gram works better than 3-gram and 4-gram in predicting
useful shortcut. Higher order N-grams require more training and are less applicable.
In the future, we plan to use mixed N-gram models as a prediction model. Multiple N-gram models
of various N used simultaneously for suggesting the best shortcuts.
It will also be interesting to mix Web content mining and Web usage mining techniques. For ex-
ample, the destination page of a session is predicted by looking at current browsing sequence as well
as its contents.
Another interesting research topic is to compare N-gram models with models that learn from their
errors such as neural networks.
r eferences
Anderson, C. R., Domingos, P., & Weld, D. S. (2001). Adaptive web navigation for wireless devices. In
Proceedings of IJCAI-01 Workshop . Seattle, WA.
Deshpande, M. & Karypis, G. (2000). Selective markov models for predicting Web-page accesses. (Tech.
Rep. No. 00-056). University of Minnesota, IN.
Frias-Martinez, E., & Karamcheti, V. (2002). A prediction model for user access sequences, In Proceed-
ings of the Workshop on Web Mining for Usage Patterns and User Profiles. .
Fu, Y., Sandhu, K., & Shih, M. (1999). Clustering of Web users based on access patterns. International
Workshop on Web Usage Analysis and User Profiling. San Diego, CA.
Koutri, M., Daskalaki, S., & Avouris N. (2002). Adaptive interaction with Web Sites: An overview of
methods and techniques. In Proceedings of the 4 th International Workshop on Computer Science and
Information Technologies .Patras, Greece.
Mobasher, B., Cooley, R., & Srivastava, J. (1999a). Automatic personalization based on web usage min-
ing. Communications of the ACM, 43 (8), 142-151.
Mobasher, B., Cooley, R. & Srivastava, J. (1999b). Data preparation for mining world wide web brows-
ing patterns. Knowledge and Information Systems, 1 (1), 5-32.