Personalised Travel Recommendation based on Location Co-occurrence

TL;DR

An extensive evaluation based on manual annotations shows that more strict ranking methods like cosine similarity and a proposed RankDiff algorithm provide more serendipitous recommendations and are able to link similar locations on opposite sides of the world.

Abstract

We propose a new task of recommending touristic locations based on a user's visiting history in a geographically remote region. This can be used to plan a touristic visit to a new city or country, or by travel agencies to provide personalised travel deals. A set of geotags is used to compute a location similarity model between two different regions. The similarity between two landmarks is derived from the number of users that have visited both places, using a Gaussian density estimation of the co-occurrence space of location visits to cluster related geotags. The standard deviation of the kernel can be used as a scale parameter that determines the size of the recommended landmarks. A personalised recommendation based on the location similarity model is evaluated on city and country scale and is able to outperform a location ranking based on popularity. Especially when a tourist filter based on visit duration is enforced, the prediction can be accurately adapted to the preference of the user. An extensive evaluation based on manual annotations shows that more strict ranking methods like cosine similarity and a proposed RankDiff algorithm provide more serendipitous recommendations and are able to link similar locations on opposite sides of the world.

Figures (14)

• Figure 1: The distribution of the number of geotagged photos per user in descending order. The accuracy filter reduces the data set from 43M to 26M geotags. By selecting only unique geotags we maintain 7M points. The table also indicates the mean and median number of geotags per user.
• Figure 2: When Flickr users make photos. Left: Photo count per week from 2003 to 2010. Right: Photo count per minute of the day, aggregated over all days.
• Figure 3: Where Flickr users make photos: World distribution.
• Figure 4: Where Flickr users make photos: USA distribution
• Figure 5: Experimental setup. The training users generate the global travel distribution $\Phi$ and the location similarity model $\Phi^{CC}$. The performance of both models for location recommendation in a predefined region $\mathcal{R}$ is evaluated on the test users.