STAS: Recommendation of Spatio-Temporal Activities Sequences

STAS : Recommandation de Séquences d'Activités Spatio-Temporelles

Diana Nurbakova (1) , Léa Laporte (1) , Sylvie Calabretto (1) , Jérôme Gensel (2)

1 DRIM - Distribution, Recherche d'Information et Mobilité
2 STEAMER - Spatio-temporal information systems

In this work, we explore a combination of users’ sequential patterns and textual and categorical information in order to provide a user with a personalised itinerary of spatio-temporal activities during a distributed event (e.g. a cruise or a festival). Firstly, we model users’ interest in activities by combining textual and categorical information. Secondly, we retrieve users' sequential patterns to model their typical transitions between activities and categories. Finally, we model the itinerary construction problem as an instance of Orienteering Problem with Time Windows (OPTW) using the obtained interest scores. We incorporate sequential patterns in the iterative algorithm to solve it. We evaluate our approach on a dataset built over a cruise program.

Mots clés: Recommender System; Itinerary construction

Domaines: Recherche d'information [cs.IR]

Diana Nurbakova, Léa Laporte, Sylvie Calabretto, Jérôme Gensel. STAS: Recommendation of Spatio-Temporal Activities Sequences. Journée des Thèses du LIRIS 2016, Nov 2016, Lyon, France. , Journée des Thèses du LIRIS 2016, pp.30-31, 2016. ⟨hal-01575736⟩