Michael Barg and Raymond K. Wong
In response to the differing relaxation needs of individual users, this paper presents novel solutions for generating improved user profiles. A method of merging stereotype information and individual profiles is suggested, so that the system can react usefully upon first encountering an individual. This paper also proposes a method whereby the system can make inferences about useful profiles to use in a given situation, in instances where the exact situation has not been previously encountered, but where similar situaions have. Finally, a method is proposed to re-evaluate the inference strategy of the system in determining the importance of the various terms for a specific user, instead of using only a single, rigid inflexible rule.