Web usage mining is a Web Mining paradigm in which Data Mining techniques are applied to Web usage data. As for Data Mining in general, the goal can be to construct a model of users’ behaviour or to directly construct an adaptive system. A potential advantage of an explicit user model is that it can be used for different purposes where an adaptive system has a specific function. For modelling user behaviour, usage mining is combined with other information about users.
Many aspects of users can be modelled:
Their interaction with a system, their interests, their knowledge, their geographical behaviour and also of course combinations of these. Modelling preferences needs information about users preferences for individual objects. This is often problematic because users are not always prepared to evaluate objects and enter the evaluations. Therefore other data are used like downloading, buying or time data.
Adaptive systems have the purpose to improve some aspect of the behaviour of the system. Improvements can be system-oriented, content-oriented (e.g. presenting information or products that relevant for the user), or business-oriented (e.g. presenting advertisements that the user is likely to buy, or that the vendor prefers to sell).
Their interaction with a system, their interests, their knowledge, their geographical behaviour and also of course combinations of these. Modelling preferences needs information about users preferences for individual objects. This is often problematic because users are not always prepared to evaluate objects and enter the evaluations. Therefore other data are used like downloading, buying or time data.
Adaptive systems have the purpose to improve some aspect of the behaviour of the system. Improvements can be system-oriented, content-oriented (e.g. presenting information or products that relevant for the user), or business-oriented (e.g. presenting advertisements that the user is likely to buy, or that the vendor prefers to sell).
Another dimension is wether the model or adaptation concerns individual users personalisation
or generic system behaviour. An intermediate form is to use usage information obtained during a session to adapt system behaviour.
or generic system behaviour. An intermediate form is to use usage information obtained during a session to adapt system behaviour.
This can also be combined with personalisation. System-oriented adaptation based on usage mining is aimed at performance optimisation, e.g. for Web servers. This is of paramount importance in large sites that incur a lot of traffic.
One of the factors leading to performance degradation is access to slow peripherals like disks, from which pages are pre-fetched upon user demand. Hence, it is of interest to devise intelligent pre-fetching mechanisms that allow for efficient caching.
The problem specification reduces to a next-event prediction, where the next event is a page fetch request, combined with an appropriate mechanism for refreshing the cash,e.g. least frequently used or most frequently used.
No comments:
Post a Comment