12.07.2015 Views

Smart User Models: Modelling the Humans in Ambient ... - UdG

Smart User Models: Modelling the Humans in Ambient ... - UdG

Smart User Models: Modelling the Humans in Ambient ... - UdG

SHOW MORE
SHOW LESS
  • No tags were found...

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

8 Gustavo González et al.5 A Case StudyIn this section, we illustrate with an example <strong>the</strong> functional operation of <strong>the</strong>architecture proposed. Let Juán Valdez be a user that has <strong>in</strong>teracted <strong>in</strong> <strong>the</strong> pastwith <strong>the</strong> IRES recommender system <strong>in</strong> <strong>the</strong> ‘restaurant doma<strong>in</strong>’ [9]. Now, JuánValdez sets up his SUM, <strong>the</strong>refore <strong>the</strong> UAA starts.In a first step, Juán Valdez <strong>in</strong>itializes his <strong>Smart</strong> <strong>User</strong> Model through <strong>the</strong> UAAby register<strong>in</strong>g his ID and his password through <strong>the</strong> Control Agent. Immediately<strong>the</strong> Control Agent requests to <strong>the</strong> Creator Agent for registration. Such latteragent, first, ga<strong>the</strong>rs <strong>the</strong> current <strong>in</strong>formation about <strong>the</strong> user <strong>in</strong> <strong>the</strong> restaurantdoma<strong>in</strong>, and sends it to <strong>the</strong> Repository Agent. Then, <strong>the</strong> Control Agent createsan Application Agent for <strong>the</strong> restaurant recommender system, and registers <strong>the</strong>restaurant application to <strong>the</strong> Control AgentAt <strong>the</strong> UAA, <strong>the</strong> Control Agent, prompts <strong>the</strong> user <strong>the</strong> <strong>in</strong>formation regard<strong>in</strong>gc<strong>in</strong>ema recommender systems and Juán Valdez selects one application. Then, <strong>the</strong>Control Agent creates an Application Agent to deal with <strong>the</strong> new application.The Application Agent looks <strong>in</strong> <strong>the</strong> social network for a user that has deal with<strong>the</strong> new application. It is <strong>the</strong> case, that Paula Allende has already <strong>in</strong>teractedwith <strong>the</strong> new application. So, <strong>the</strong> correspond<strong>in</strong>g Application Agent of Paula andJuan dialog. The Application Agent of Juán acquires <strong>the</strong> graph G (SUM, UMD i )correspond<strong>in</strong>g to <strong>the</strong> relationship between <strong>the</strong> SUM of Paula and her UMD <strong>in</strong><strong>the</strong> ‘c<strong>in</strong>ema doma<strong>in</strong>’. Weighted graph G is adapted to <strong>the</strong> SUM of Juán Valdezand his Application Agent is ready to deal with <strong>the</strong> recommendation process.After a while, <strong>the</strong> WSAA Provider Agent ga<strong>the</strong>rs <strong>in</strong>formation about newrecommender systems <strong>in</strong> <strong>the</strong> ‘restaurant doma<strong>in</strong>’. That is <strong>the</strong> case, Juán Valdezhas used until now a recommender system of <strong>the</strong> Girona city, and <strong>the</strong> ProviderAgent has discovered a recommender system about <strong>the</strong> Barcelona city. S<strong>in</strong>ce <strong>the</strong>user is travell<strong>in</strong>g round this latter locality (contextual <strong>in</strong>formation), <strong>the</strong> ProviderAgent believes that such <strong>in</strong>formation can be <strong>in</strong>terest<strong>in</strong>g for <strong>the</strong> user. Hence, <strong>the</strong>WSAA requests to <strong>the</strong> UAA about <strong>the</strong> possibility of generat<strong>in</strong>g a new applicationon this new recommender system.6 ConclusionsIn <strong>the</strong> past, user modell<strong>in</strong>g had focused its attention <strong>in</strong> develop<strong>in</strong>g doma<strong>in</strong>dependentsoftware architectures for user models [10,11,12,13]. However, <strong>in</strong> recentyears, <strong>in</strong>formation technology has moved from s<strong>in</strong>gle and centralized usesto distributed multipurpose systems, which are now <strong>in</strong>creas<strong>in</strong>gly embedded <strong>in</strong> afully <strong>in</strong>terconnected world [14,15]. The <strong>Smart</strong> <strong>User</strong> <strong>Models</strong> not only must haverelation with <strong>the</strong> context <strong>in</strong> where <strong>the</strong>se are used but also <strong>the</strong>se are a cornerstone<strong>in</strong> cross-doma<strong>in</strong> recommendation processes. For <strong>in</strong>stance, if a user modelhelps to a user to choose a comfortable restaurant, this same user model couldhelp him/her to choose a comfortable c<strong>in</strong>ema, although <strong>the</strong> doma<strong>in</strong>s are different.Therefore, we are contribut<strong>in</strong>g to <strong>the</strong> next generation of open environmentsthrough <strong>Smart</strong> <strong>User</strong> <strong>Models</strong> which <strong>in</strong>clude between o<strong>the</strong>rs <strong>the</strong> emotional factorof <strong>the</strong> user, which <strong>the</strong>y represent.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!