This paper aims to make a proposal for overcoming some of the user preference capture process limitations, by means of a mechanism of “Assisted Browsing” for the initial moments of the user interaction. The approach we present introduces a mechanism for users to browse, in a guided way (what we call “assisted browsing”), among subsets of the real content so that they can express their preferences more effectively, and at the same time, the system can capture data related to actual user decisions (“Yes, more like this”, “No, take me elsewhere”) with respect to content, instead or mere opinions about the meaning of the categories. The concept of “Assisted Browsing” has been taken from some systems appeared in late 90s and which can be found in some services for providing catalogues in e-commerce. However, our approach uses some procedures coming from Psychometrics computerized tests (or CAT, Computer Assisted Testing), which are new in the personalisation field. This paper reports on the ongoing research, including some details on the initial specification of the algorithms.
More...