| dc.description.abstract | Context-Aware Computing describes the development of technologies and applications that can detect data from the surrounding environment and react accordingly with specific actions, reducing and simplifying the human-machine interaction process. The latter automatically offer a range of services to help the user during daily professional or private life by managing the available resources. Therefore, Context Awareness (CA) should be intended as a set of technical features able to provide added value to services in different application segments.
Context changes result in a transformation of the user experience. For this reason, Context-Aware Computing has played an essential role in addressing this challenge in previous paradigms, such as Mobile and Pervasive Computing, and is playing a crucial role in the new Internet of Things (IoT) paradigm.
Over the last years, e-Tourism and, in particular, Cultural Heritage have provided two main domains for this type of research. Indeed, thanks to new technologies, a tourist can access large amounts of contents and services before, during and after visiting experience, with different purposes and requirements in each phase. In this scenario, the need arises for Recommendation Systems (RS) that consider users' personal preferences and all the contextual aspects to recommend the right services and contents at a specific time.
The research activity concerned the study of Context-Aware Recommender Systems (CARS), focusing on the modelling and managing all the possible Contexts in an application domain. In particular, the problem of tailored data modelling has arisen, as this represents an enabler for new information systems: Mobile Systems, Big Data Systems, P2P Systems and, in general, the Semantic Web.
In that regard, a system architecture for the fruition of e-Tourism contents and services was designed to enhance the Cultural Heritage by responding in a unique way to the Context and users' needs. It will be capable of supporting not only visiting users but also public institutions and sector operators through the automatic as well as the dynamic definition and recommendation of core and ancillary services for tourism promotion: from the search for a destination up to the use of Cultural Heritage-related content and the commentary of the visitor experience, including tourism promotion services, booking, e-ticketing, e-commerce, social networking.
To recommend contextual contents and services, the innovative characteristics of the proposed approach mainly concern the information to be made available to end-users, suggesting three main points of view:
1. Data Management and Inferential Engines
In such a scenario, data represent the key to build up and enable services and actions to take: the goal is to implement a Knowledge Base (KB) to collect, elaborate, and manage information in real-time. In this respect, Knowledge Organization Systems (KOS) refer to well-known schemes such as taxonomies, thesauri and other types of vocabularies that, together with ontologies, constitute valuable tools to shape the reality of interest into concepts and relations between concepts.
The system, thought to be continuously functioning, collects data from various sources without interruption and immediately processes them, intending to activate precise actions, depending on the users and the events. The latter, detected and analysed, will have to be translated into facts associated with specific semantic values: it is necessary to use inferential engines capable of drawing some conclusions by applying particular rules to reported facts. In this regard, many approaches are based on the so-called Bayesian Networks: powerful conceptual, mathematic and application tools allowing the management of complex problems with a significant number of variables interlinked by both probabilistic and deterministic relations. Such networks also make it possible to update the probabilities of all the variables involved whenever new information is collected on some of them, using Bayes' theorem.
2. Context Representation
The goal is primarily to deliver to different categories of users, in each moment, information that is useful in a given Context. In practice, the objective would be setting up an architecture characterised by a high degree of Context-Awareness. Real-time understanding of the Context where users are located, via a representation by means of graphs, allows indeed to provide them with a wide array of "tailored" services and hints regarding the decisions to make, managing in the best possible way both the time and resources they have and showing them what is around, ultimately meeting their needs. More in detail, the Context's representation can be implemented through formal models of representation, such as the Context Dimension Tree (CDT).
3. Recommender Systems
A Context-Aware System's ability to reduce information noise takes on considerable importance together with the possibility of the system itself generating an ordered list of personalised suggestions in each Context through a recommendation engine. Recommender Systems are applied in different sectors but have one goal: to help people make choices based on an analysis of users and items in terms of main features. In other words, the purpose is to predict the consideration that an individual may have about an object that he has not yet evaluated.
Ultimately, the goal is to identify a framework, mainly based on a powerful contextual recommendation engine, to be a highly flexible inferential and decision-making tool. This framework does not only allow to manage of complex problems, featuring a great variety of variables inter-linked through both logical-deterministic and probabilistic relationships, but it also provides an adequate representation of the phenomenon at stake. In fact, it simplifies the problem description as well as the summary easier, enhancing the degree of its comprehension and allowing to identify the key variables. In addition, modularity allows for easy integration of new functionalities that can be developed and tested separately, such as a process capable of presenting information in learning environments according to Digital Storytelling techniques.
Based on the proposed architecture, an application prototype was developed to support the user in the construction of a personalised and contextualised tourist route related to some of the most important cultural sites in Campania (a region in Southern Italy): a hybrid mobile application designed and implemented together with a server-side component.
The experimental results show the ability of the system to be effective. Future activities include improving the developed prototype, including a chatbot, and an experimental campaign involving a more significant number of users. [edited by Author] | it_IT |
