|
User Interaction
ICE provides a number of methods for users to discover the most relevant
content from each of these areas:
Natural Language Search
ICE allows a user to enter a query in full natural language rather than the older
keyword and boolean approaches. Results are provided with intelligent query-specific
summaries.
Automatic Profiling
Once a user is identified, ICE reads along with them, learning the subjects that
interest the user. Each user's visit to the site is then highly personalised with
no effort on their part. As this profile is generated automatically and in real
time it is always up-to-date, and reflects the variety of interests that a user
may have.
Trackers
ICE allows a user to enter their interests in natural language. This allows a
user to keep up to date with the subjects that interest them automatically. Rather
than most present methods which rely on a user selecting very broad subjects from
a list, ICE allows users to specify very clearly the subject that interests them.
Not 'rowing' but 'Steve Redgrave', not 'Athletics' but '100m final'. ICE will
also look across the complete content for relevant articles and looks at the article
concepts, not the keyword tags that most present systems use.
Path of Interest
Because ICE understands which concepts are more important than others within an
article it is able to automatically link related information together. Rather
than a publisher having to search for relevant related articles, and categorise
that into a very limited set of subjects, ICE automatically and dynamically provides
much richer related links across the complete range of concepts contained within
an article.
Community Development
ICE builds up an understanding of a user's interests, by watching the articles
they read, by noting the queries submitted, and by noting the tracker definitions.
Because of this, ICE has a comprehensive understanding of the interests of that
user, and their relative importances. From this, ICE is able to look at not only
linking users to articles, but also linking users directly with other relevant
users.
|
