A browsing process is driven by user interaction. The user is presented with a selection of items 511, 512, . . . etc from a range. The user can give inputs representing interest in one or other of the items displayed. The inputs represent rewards, which are distributed to attribute data items or “keywords” associated with the display items. These keywords are transparent to the user, but represent characteristics of the display items with which they are associated. The browsing system selects items for display according to a probabilistic function weighted to favour those having the keywords which have accrued the highest number of rewards, these being the display items having the most keywords in common with display items previously rewarded by the user during the browsing session. Partial or negative weightings may also be applied to the keyword associations, which are also taken into account in the selection process.