Ricardo Alonso Maturana, CEO of Gnoss and John Mora Williams, CEO of Globalditauthors of the article; they are also working as Directors of the project Ontological Conceptual Model of the Tourism Domain commissioned by SEGITTUR.
In most organisations, industries and markets there is a growing disconnect between machine learning algorithms and corporate knowledge. This disconnect is an example of a more general problem in the Artificial Intelligence (IA). As pointed out by DARPA (Defense Advanced Research Projects Agency) the US Defense Advanced Research Projects Agency, when machine learning algorithms do not incorporate relevant domain knowledge, whether in banking, culture, education, research, automated driving, defence, medical diagnosis, accurate description of a tourist destination or any other sector, they can generate false results that, depending on the criticality of the decisions involved, can be catastrophic. Recently, Craig Martell, the Pentagon's Chief of Artificial and Digital Intelligence, has stated in a interview with CNN who is "scared to death" of generative AI. He points out that, while large linguistic models are unreliable, "these things speak with authority, so we believe them. That scares me.
It is therefore essential to connect algorithms and data processing with rich domain knowledge ready to reason and make inferences. The kind of technical artefacts capable of making such connections are known as ontologies. An ontology projects onto machines the cognitive structure by which people understand the world in a given domain of knowledge, so that they can understand and process it as we do. Ontologies are thus models of knowledge representation independent of any particular representation of databases, files, systems, or any other means of data storage and management. Leading companies such as Amazon, Google, Facebook, Wal-Mart or Nike are developing ontological programmes based on knowledge graphs, precisely because they want to have an artificially constructed intelligence that is explainable and whose results are traceable, i.e. an AI that is connected to what we commonly understand as "truth". Machines and people need to converse in a framework of "common sense" and knowledge graphs, which represent the entities of an ontologically represented knowledge domain, are the means to do so.
There is a growing interest in ontologies and knowledge graphs in all sectors.
Increasingly, forward-looking companies and sectors, those that understand the importance of knowledge, its traceability and therefore its reliability, are developing AI master plans that contain semantics at their core. This is the case of tourism in Spain with the Tourism Ontological Model-promoted from SEGITTURwhich lays the foundations for a national project of Tourism Intelligence. The word "infostructure" is a neologism that serves to designate a basic information infrastructure, which in the field of knowledge would function like the network of ports and roads in the field of logistics or the transport and distribution of goods or passengers. Following this analogy, the Ontological Model Tourism-Spain can be considered a National Knowledge Infostructure.
Having an ontology, an ontological model, is the condition for building intelligent machines and systems that satisfy the conditions that the latest Gartner report in 2022 calls Composite AI, which is the one that hybridizes the semantic technologies and the knowledge networksThis ensures the explainability and traceability of their results, with heuristic and generative technologies, which make it possible to "humanise" the relationship between people and machines. The Tourism-Spain Ontological Modelis, therefore, the condition for building Composite AI solutions in destinations that satisfy the conditions of what we can call the European Way of AI, strongly committed to the reliability of the results. Knowledge ontologies and knowledge graphs allow to endow Meaning, Reasoning and Learning systems (Meaning, Reasoning y Learning), the three conditions that a Strong, Composed and Reliable AI must satisfy.
Tourism is Conversations.
Tourism is about conversations between people, but also between machines and people.
A smart territory is one that is able to interpret what is happening and proactively converse with travellers before, during and after their visit.
Tourism is about personal conversations, because each traveller is different, wants different things and aspires to different experiences. The Tourism-Spain Ontological Model is the condition for systems to be able to talk to everyone, to whisper in everyone's ear what they need to know at a precise time and place, and to advise them on what is most appropriate, best to do or to take into account. In short, it is the condition for machines to be able in the future to talk with sense, opportunity and usefulness with every person who approaches or lives in any part of Spain. The Tourism-Spain Ontological Model will enable the future Intelligent Destination Platform of Spain (promoted by SEGITTUR) to enable technological capabilities to expand, make the tourism exploitation model of our country more sustainable and useful for tourists, to transform it digitally. This common "vocabulary" will favour interoperability between different systems in the destination, between destinations and, finally, between these and the central node of the smart destination platform. For the Smart DestinationsThis instrument is set to be one of the accelerators of their competitiveness with the greatest impact, as it makes it easier for their value proposition to be "understood" by machines, which will guarantee the delivery of personalised services to travellers, collaboration with other destinations for the resolution of different tourism use cases, the activation of enriched services through composite AI or the sharing of data with public and private operators.
The Ontological Model of Tourism-Spain is not only a Conceptual Reference Model, but a machine-operable ontology that contains all the classes, attributes, relations, extensions and use cases that can be deployed by the whole of the Spanish tourism offer, as well as the modes of interaction of the different types of travellers. The expression "operable ontology" refers to the fact that the ontological model has been written in a technical programming language interpretable by machines, OWL, Ontology Web Language, the standard language defined by W3C for programming ontologies, which ensures that each concept is framed in an unambiguous way and that it is readable by both humans and machines.
The Ontological Model of Tourism-Spain
It is worth considering some of the background to the Tourism Ontology Spain. The team of GNOSS developed in 2013 and part of 2014 the institutional web portal for tourism of the Government of La Riojawith the impetus of the then Minister of Education, Culture and Tourism and current President of La Rioja, Gonzalo Capellán. La Rioja Turismo" exploits a tourist Knowledge Graph in which all the digital contents referring to places of interest, accommodation, routes, activities, events, restaurants, wineries, etc., are represented using an ontology written in RDF/OWL. The construction of the Semantic Ontology Graph hybridised and extended several pre-existing domain ontologies and vocabularies with the aim of generating a digital space where the experience of retrieving information and, in general, conversing with the machine would be simpler, more useful and practical, and where the overall web experience would be friendlier and more satisfying. The ontologies that were then hybridised were Harmonise, OnTour, Geonames, Rout, FRBR and rNews. Currently, the Knowledge Graph of ?La Rioja Tourism? consists of more than 7,000 digital contents, 67,284 entities, 472,361 relations and 675,368 triples. The most significant exploitations of the Knowledge Network are linked to the existence of a meta-search engine, faceted search engines for each knowledge object, information systems contextual and display systems of the Graph by combining the map with semantic geopositioning.
La Rioja Turismo was launched in June 2014 and since then it has functioned as the tourism publication space for La Rioja. Currently, in Spain, it is still the only destination platform that works in this intelligent way and that would eventually make it possible to deploy a semantic services space for all the actors in the sector. The project was presented at the 1st International Workshop on Knowledge Graphs on Travel and Tourism. Full-day Workshop at the 18th International Conference on Web Engineering (ICWE 2018).
On the other hand, the CTN 178 or Technical Committee for Standardisation 178 of the Spanish Association for Standardisation or UNE is a public-private collaboration group with more than 700 experts that works on smart cities and which in Spain is the standardisation reference framework for everything related to processes, technologies and operations of this kind of cities. In CTN178 of UNE, in 2018, the analysis, drafting and consensus work on the UNE 178503 standard, Semantics applied to smart cities, was completed. Smart Destinationswhich can be considered an intellectual predecessor of the Tourism Ontology-Spain. The resulting document offered the sector a first vocabulary that allowed, and still allows, to describe its operations thanks to definitions, taxonomy and a proposal for data exchange based on the JSON format. The standard established equivalences between the proposed vocabulary and SCHEMA.ORG, which was considered to function as a guide for developers. At the time of COVID-19, the WG7 in charge of the standard published an update containing terms and guidelines for operating critical destination information in extraordinary circumstances, such as those.
The aim of the design of the Conceptual Reference Model of the tourism domain was to define a semantic model aligned with the UNE 178503 Standard, which can be used as a guide and basis for the construction of a Network of Ontologies that digitally reflects the context of tourism consumption. In order to build the Ontological Model of Tourism-Spain, which will operate at the service of the future Intelligent Destination Platform (IDP) promoted by SEGITTUR, a "Conceptual Semantic Reference Model of Tourism" has been previously developed.
The definition of the model considered as a starting point the subdomains of Supply, Tourist, and Destination, connecting all of these with the concepts of tourist interaction and addressing the travel cycle (before, during and after).
On that basis, the ontology comprises 305 classes, 433 properties and 103 lists of terms in the defined and already mentioned subdomains of offer, tourist, destination. Of the 305 classes, 13 of them (Accommodation, Tourist guide, Tourist or related facility, Tourist intermediary, Local business, Event organisation, Catering, Passenger transport, Tourist destination, Transport infrastructure, Resource of tourist interest, Public service and Person) have been defined as "main classes" from the business point of view. It is, as can be seen, a model designed with a high level of ambition from the point of view of the expressiveness of the representation of what the sector does or can do.
In short, the Tourism Ontology-Spain aims to become a living, highly expressive, evolving and extensible standard, enabling the different actors in the tourism market to talk to each other: people to machines, machines to people and machines to each other. In other words, it will enable people to talk to destinations and destinations to talk to people.
The Model will thus contribute to building new experiences for travellers in the destination, to accelerating the transformation of the processes that together contribute to building the tourism offer and to opening up the sector to new digital businesses. digital transformation of tourism in Spain.
Finally, the model is destined to be the central element of semantic interoperability of the set of destinations that make up the aggregate offer of destination Spain, which will enable its unification from a data point of view. Based on this unified reality from a data point of view, it will be possible in the future to generate all kinds of cross-cutting tourist stories or narratives which will help to facilitate the relocation and deseasonalisation of tourist consumption, a strategic objective which would help to moderate the saturation of the most used tourist areas, as well as adding value to others which are currently devoid of almost any economic activity.
