About sightsplanner

Sightsplanner is a personal recommender for your trip to a foreign city. It was launched for Tallinn in spring 2011.

Based on your time of the visit and your preferences about different types of sights and events, the system will suggest which places to visit and will build a realistic itinerary just for your visit.

Why use sightsplanner?

Obviously, most people planning a trip start investigations by looking for materials on the web. It is easy to find a large amount of information about the city and its most popular tourism objects, but it takes very long time to select and organize all the interesting objects. Places and events which are really interesting for you, but not for the general public, are easily missed.

Once you find the sights you'd like to see, you need to build a rough itinerary. This is - obviously - pretty hard. We need to find out opening times and schedule the visits accordingly. It may turn out that some of the objects do not fit into the time schedule, in which we will have need to select different objects. All this takes a lot of effort.

Sightsplanner takes care of finding the objects based on you personal preferences, preferences, visit time, method of travel and the corresponding information about sights: their types, opening times, location, typical visit time and popularity of the sight. It will also build an itinerary for your visit. Last not least, sightsplanner knows more objects and more about the objects than any other site.

Using sightsplanner

Your plan and suggestions are composed using your preferences, visit time, method of travel and the corresponding information about sights: their types, opening times, location, typical visit time and popularity of the sight.

Set your preferences by moving sliders left or right. Moving the slider left will give fewer suggestions in that category, moving right will give you more. You can set more detailed preferences by opening the subcategory list: just click on the category name and the detailed list will open.

After setting your preferences, click the "Look" button: this will build your personalized visit itinerary. If you just want to see the most popular objects of the selected category mix, click the link "Popular objects" below the button. At the bottom of the page there are several options for defining your visit - start time and date, duration and means of transport. The question marks after the sliders give a few hints about what might be included in that category. The four pre-set options at the top of the page simply move several sliders at once to a predefined position. 

How does it work?

The suggestion engine finds interesting objects for the given user, based on the user preferences, using sophisticated probabilistic matching algorithms and a wealth of knowledge about the objects. For example, sightsplanner knows the rough popularity of the object, it knows the different categories (along with fuzzy degrees) the object is in, it knows the average visit time and opening times. Each knowledge item is associated with a degree of confidence.

In order to find the interesting objects, each query recalculates the score of "interestingness" for absolutely all the objects in the database, for exactly the user-given set of preferences and limitations of the given query. Simply put, the objects that the user probably likes have a higher score and vice versa.

The objects and events with their calculates "interestingness" are then organized into a timetable based on their location and time using several optimal trip search algorithms in parallel.

Conceptually, we use extended semantic web techonology plus probabilistic rules with a confidence measure. Data is kept as extended RDF sextuplets - object, property, value, confidence, source, timestamp - in both an ordinary relational database and an extremely fast main memory database for match calculation and planning. We derive additional information for objects using a rule system including full predicate logic augmented with a confidence measure.

The system database is built by merging, scraping and converting information from a wide number of different sources: several structured databases, several semistructured thematic websites plus human research and data augmentation/correction.

Contacts and links

Sightsplanner was built by a team of developers and researchers from three companies Mindstone, Apprise, ELIKO and the Tallinn University of Technology.

For further information, contact Tanel Tammet: tanel.tammet@gmail.com

Some links you may want to check:

• Sightsplanner system
• Facebook page
  
• Blog