• Better customer satisfaction
• Shorter session time

The PeToMoTo project is funded by the German Federal Ministry of Economics and Technology on the basis of a decision by the German Bundestag. The project has a cooperation agreement with Alpstein Tourismus GmbH & Co. KG for providing real application data and its expertise of the tourism industry.

Users searching for tour recommendations in the outdoor tourism are often disappointed by the behavior of contemporary recommender systems: There might be too many recommendations or quite the contrary not any recommendation. A mobile outdoor scenario even sharpens users' dilemma when trying to get rid of the implications of critical weather changes for the worse. Complex spatial objects and dynamically changing situations claim new approaches. The growing social nets give also hints how to profit from users' data by modeling their data as preferences. Thus, preference research based on data bases is an adequate starting point to model users' wishes and dislikes accordingly.

The scientific core of the PeToMoTo project is described by a consistent extension of the preference model and its connotated situation model with respect to the specific requirements of a mobile outdoor scenario. All innovations have been integrated in the already existing 'Preference SQL' system which is the base in order to construct individual preference-based queries for outdoor activities. The implemented recommender automatically takes care of the users' input, their selected roles, and the weather conditions depending of the selected location. It only delivers the best tours according to the specified conditions and preferences. Just one query is implicitly generated after having clicked the action button. Thus ‘preference search’ is a '1-click search' indeed. No further activities like multidimensional parameter variations are necessary pruning an annoying process of trial and error obviously.

A test version of the commercially used outdooractive.com portal of Alpstein has been implemented by the Preference SQL framework. The preference search prototype was benchmarked by a small group with the running portal implementing faceted search. The comparison unveils higher customer satisfaction and shorter session times by using the preference search prototype.

Fig. 1. Man machine interface of PeToMoTo for preference search

In Fig. 1, a user (e.g. Sue) decides which activity (e.g. hiking) she likes in her favorite region like Allgäu. Then, she adjusts to one of predefined roles as tourist. The length of her dream tour is chosen to be approximately 9.0 km, the ascent to be approximately 300 m, and the duration to be approximately 4 h. The wished tours can have any difficulty degree. Since the region is known, its current weather state is requested by a weather API. The situation model derives appropriate preferences with regard to e.g. the type of activity, the minimal and maximal height. Knowing her input, her role, and the current weather the preference generator consistently mixes all individual input sources to formulate just a single Preference SQL statement, which encodes Sue's constraints and preferences. After pushing the search button, the best recommendations are only presented. Normally, the result set consists of up to 15 recommended tours which are equally good with respect to the users' specification. Worse tours are automatically canceled.

Start date:   01.11.2010

End date:    31.10.2011

Funded by: MWi (Bundesministerium für Wirtschaft und Technologie)

Local head of project: Prof. Dr. Werner Kießling

Local scientists:

Dr. Alfons Huhn
Patrick Roocks
Andreas Zelend
Priv.-Doz. Dr. Markus Endres

External scientists / cooperations:Alpstein Tourismus GmbH & Co. KG