Information Technology Reference
In-Depth Information
adaptive sequencing. The difference between direct guidance and ITS adap-
tive sequencing disappears gradually in the Web context. As long as some
types of educational material, such as presentations and questions, are rep-
resented as a set of nodes in hyperspace, the sequencing becomes indistin-
guishable from direct guidance. Popular examples are ELM-ART [14] and
InterBook [13].
17.3 Personalization Approach
The goal of personalization is to provide users with what they want or
need without asking explicitly [39]. In this section, different approaches are
reviewed on how the personalization can be delivered to individual users.
17.3.1 individual versus Collaborative
A personalization system may be built based on an individual user
profile to predict and tailor future interactions. It is called the individ-
ual approach . This approach requires content descriptions and is often
referred to as content-based filtering systems. NewsWeeder [32] is an
example of using content-based filtering that automatically learns user
profiles to recommend articles to the user. The major disadvantage of this
approach is the filtering only relies on the users' previous interests for the
recommendation.
There is an alternative approach that not only uses the individual user pro-
file, but also takes care of other users who share similar preferences. It is called
the collaborative approach . This approach is referred to as collaborative filter-
ing. GroupLens [45] is an example that recommends articles to users based on
a similar user profile. The major disadvantage of this approach is the reliance
on the availability of ratings for any item prior to recommending it.
17.3.2 reactive versus Proactive
The reactive approach to personalization refers to a conversational process
that requires explicit interactions with the user in the form of queries or
feedback. To provide the items of interest to the user, the feedback is also
incorporated into the recommendation process for refining the search and
suggestion. Most reactive systems for personalization have their origins in
case-based reasoning research [16,34,35]. Reactive systems can also be clas-
sified based on the two major feedbacks: common feedback [34] and prefer-
ence feedback [35]. In common feedback, the user must provide a rating for
each recommendation based on the suitability to the user's needs. In prefer-
ence feedback, the user is provided with a list of recommendations based on
 
Search WWH ::




Custom Search