How Carta works

Data assembly

We gather and align data from (open) (travel) encyclopedias, topical databases, partners, our community and more.

Place partonomy

The characteristics of a region are the result of the aggregated characteristics of all the destinations it contains. For example, a region with a lot of mountain bike routes can only be compared with other regions based on the sum of the routes they contain. In many cases places only partly overlap. For example, a route may be streched over multiple regions. And some destinations move from place to place, like a musician. We transfer characteristics between destinations based on the degree of overlap and time. In this way we can recommend destinations for things they aren't explicitly described.

Extraction

As pointed out, we work with structured sources, but we also mine texts with full sentences: 'natural language'. We identify relevant concepts in them with our natural language miner.

Ontology

The above described concept extraction is focused on terms that may be relevant to travel or leisure. We store relevant concepts and connect them amongst others with taxonomic relationships. For example, if a place is known for tennis, football, or cycling, we can derive that it is also know for sports in general. This kind of knowledge signficantly improves recommendations.

Experiences, activity & plans

A place can have a beautiful description, be tagged for various interesting things, but if it's not liked by others, it shouldn't be recommended. So, we assemble reviews and experiences. These opinions don't just say something about the thing they review, but also about the interests and values of the reviewers themselves.

Relations

Not everyone's input is probably equally important to you in daily life. The opinion of some people may be very influential among their friends or followers, but perhaps not to you. We derive whether people are close in terms of interests, activities, opinions, and friend-relations. It's as if we extend your friend group with others you haven't yet met. In this way, we can generate personalized recommendations for destinations on the other side of the globe. Destinations that none of your actual friends have visited.

Your quest

You indicate what you're looking for. What kind of things: Currently we cover national parks, but also think of countries and regions or things like museums, campings, routes, or beaches. What they should be known for: This can be anything. Perhaps friendly locals or staff, or a certain difficulty for your sport route, or any combination of things. And where: Search the whole world, check a certain region, or zoom in to a neighborhood.

Hybrid recommendation

We check which places have what you're looking for, whether they are liked and by whom. We generate a score based on this. We also take things like weather and (other) seasonal changes into account. We highlight a small selection of destinations, making a good decision much easier.

On the map

We show the recommendations on the map. In this way you can quickly see how accessible they are in relation to you or other places that you may want to visit. You also get an image of the topographic features.