All articles

En route to graphs

The platform of Outdooractive is constantly moving forward. Thus, the Outdooractive platform is working on the next generation of data structures.

But first things first. Buzzwords such as graph database, Knowledge Graph or linked open data are more and more in the center of focus for digital tourism professionals. This shows that it is time to take a serious look data and their structures.

What do the terms mean? Sector experts may forgive me for the following explanation that even non-specialists can understand:

The aim is to create a data structure in which all touristic objects are linked to each other. In tourism, unlike other industries, there is a huge amount of diverse information and countless contributors who create content. In the world of sustainable databases data is displayed in form of objects. First and foremost, objects are the different types of data (e.g. hotels, POIs, events, tours, etc.). When linking said objects, pictures are assigned to accommodations, points of refreshment/rest to tours, events to event locations (POI), bookable ski passes to a ski resorts, and so forth.

Within the individual types of objects there are also structures that ideally exist of many individual data input fields (e.g. title, summary, text, directions, geo reference, etc.). At first, these input fields are simply part of the data set of the object. Due to the structured acquisition of the individual parts of an object, they can also be broken down into separate objects and linked with other objects. For example, an image assigned to a POI can inherit information from a POI via the linkage, e.g. the geo-reference, resulting in a geo-referenced image. But also the title of a POI (e.g. Golden Deer) and the category (guesthouse) can be passed on with the image. Thus, the image has both terms as tags (buzzwords). A search for “Guesthouse Golden Deer” would now find the exact image without these terms being specifically assigned to the image. This can be continued endlessly. For example, due to geo-referencing the guest knows in which community the guesthouse is located or the touristic region, the mountain range, etc. Next to the characteristics of the guesthouse, the image also knows what destination it displays. Thus, you can also find it if you search for “destination in the Franconian forest”.

All these objects might derive from different sources. Meaning, image and POI are from different people just like the region with all its content is also from someone else, etc. The system assembles the content from different sources to provide the best possible information for the user. This is called a Knowledge Graph.

Google, for example, provides result pages that are pieced together from different sources (without naming the sources). These compiled results often appear in a separate column on the right.

From a technical point of view, a graph is first and foremost a node-edge-model. The individual elements are the nodes and the connections between them are edges. Routing is the easiest way to imagine this. The crossroads are nodes whereas streets and paths between them are edges. Next to the connection, edges can also contain additional information. For example, the connection between two objects is an asphalted country road with a width of 6 m and little traffic. The connection can also be geographical (located in a nature reserve), a condition (closed for trucks), an opening time (closed in winter), a status (now open) or a price (8 € toll). This is also called a semantic data model.

If you also make clear in the data structure – in this case via the connection with the category – that in this case, for example, the golden deer is not an animal but an inn, the basis is created for the algorithms to interpret this correctly for all displays and data transfers. This is called a semantic web.

In case you want to dig into the topic even further, you should take a look at the respective Wikipedia entry: Wikipedia itself is a prominent example of data storage with a high degree of linkage – but with relatively little logical structure.

A big data network

In the end, the entire graph-based touristic database is an immense network of individual objects linked by intelligent connections. A user can click for hours on end in such a data network without having to leave the system. This is precisely why this kind of data structure is also suitable as a basis for all language assistants. You can ask complex questions and the algorithm of an artificial assistant can draw an answer from the logically organized data structure.

How does this work at Outdooractive?

We at Outdooractive started building a digital platform 15 years ago. True to our engineering origins, we started at the very bottom and build our way up to a client-capable database in which everyone involved in tourism can participate. At a time when all other companies in the market were busy building websites with colorful images and long texts, we were mainly concerned with the data management and data structures of the future. We have modelled out all objects occurring in tourism with all their characteristics. For the creation of intelligent relationships between objects we have taught the Outdooractive software system many automatic actions.

When saving a POI, 15 to 20 different relations are created such as the region, the state, the respective nature reserve or nearby waters. But there are also connections that are (for now) done editorially like a fitting choice of rest/refreshment points of a tour. However, even this manual categorization is already supported by machines today and can be completely omitted in the future once the machine results have reached an equivalent quality.

Innovators of the touristic market (destinations, consultants and software businesses) currently discuss everything regarding collaborative databases, Knowledge Graph and linked (open) data. Thus, the time has come for Outdooractive clients to profit from many years of development. In all modesty, we at Outdooractive can proudly say: We have had a linked database for a long time. The technology and a continuous use takes the Outdooractive clients a huge step forward. But of course this does not mean that we are done with our development. On the contrary, this is just the beginning. At Outdooractive we are constantly developing the database technology; next en route to graphs and whatever else may come in the future. We are happy to take part in all current discussions about future tourist databases as well as contribute and take on new tasks for our own development. Our aim at this point is quite clear: We want to continue to provide our customers with the most powerful system and further develop it with them.