Recommendation Based on Opportunistic Information Sharing Between Tourists

Abstract

We propose a new approach to collaborative filtering in mobile tourist information systems based on spatio-temporal proximity in social contexts. The approach is motivated by a survey of festival visitors confirming that similarities of interests extends beyond events defining specific social contexts. We show how opportunistic information sharing in mobile ad-hoc networks can be used to realise decentralised collaborative filtering appropriate for mobile environments and show its equivalence to existing centralised approaches.

Information Sharing Modalities for Mobile Ad-Hoc Networks

“…Further, they have installed an application on their mobile phone that allows them to take advantage of the new technical facilities that this particular theatre offers to registered customers. Equipped with their mobile phone, our customer heads to the theatre (1). As soon as the user enters the building, the phone connects to a theatre server and the current movie schedule is displayed on their device (2).…”

“…Thus, the very way in which data is shared filters that data according to user similarity and, therefore, the first step of user-based filtering to compute similar users is not required. In [1], we also report on studies carried out to validate the underlying assumption. Note that if users have exchanged ratings and then edit them afterwards when they are no longer connected, then the updates should be propagated if and when the users encounter each other again.…”

“…Both forms of information sharing have been used in different ways in a variety of applications. For example, adhoc network connectivity between peers has been used to detect the physical copresence of users in social settings as a basis for exchanging rating data in a P2P manner in recommender systems [1]. However, P2P connections between mobile devices have also been used as a means for storing data associated with a location based on the movement of data between peers as they pass through that location [2].…”

“…In previous work [1], we have shown how ad-hoc connectivity can be used to adapt collaborative filtering (CF) algorithms to mobile settings. Typical CF algorithms such as user-based filtering proceed in two steps, first selecting a set of similar users and then aggregating the ratings made by these users.…”

A Platform to Support Object Database Research.

“…Moreover, based on the general collaboration concepts introduced by the shared collection module, different distribution models and modes of information sharing can be realised by means of configuration rather than implementation. As a result, we were able to investigate a number of different application scenarios including a recommender system where spatio-temporal proximity was used as the basis for user similarity in collaborative filtering [dSNG08].…”