Public digital displays could greatly benefit fromthe ability to dynamically select from the Internet contentitems that would be strongly related with the place where eachdisplay is installed. Generically, this is similar to the type ofproblem addressed by recommender systems. However, theusage context of a public display raises specific challenges thatmay limit the applicability of existing recommender systems.In this paper, we explore the creation of a recommendersystem for public situated displays that is able to autonomouslyselect relevant content from Internet sources using keywordsas input. This type of recommender system should enablepublic displays to become devices for Internet informationdelivery in public spaces, while also making them moresituated in the social settings in which they are installed. Wehave created a recommender system based on these principlesand we have conducted two studies to evaluate the perceivedperformance of the system. The results have shown thatkeywords can be very effective in driving user-generatedcontent, but they often need to be complemented withcontextual information that disambiguates their semantics
展开▼
