Detailed description of project

The tourist experience is a multidimensional process of finding answers to a series of questions: Where to go? What sights would it be worth a visit? Where should I stay? Where should I have lunch? How am I supposed to have fun? How do I get to an attraction or recreation area? The answers should be aimed at entertaining the visitor, taking into account his preferences, drawing on appropriate information from a huge volume of subjective experiences, such as those recorded on social media or collected by interconnected devices in the age of the Internet of Things. Subjective experiences, such as recorded in videos, photos, songs, tags, texts, ratings, geo-information trails, user history (profiles) are indeed important data (big data) both in terms of volume, diversity, speed of renewal, but also added value for the operators of the tourism industry, as well as for individual visitors.
The huge amount of information creates a scientific and technological challenge: to develop innovative tools and applications that indicate (recommend) personalized and targeted tourist information to visitors using data from the above mentioned multiple sources. The basic principle of the project is that the personalized and targeted tourist information that will be indicated to visitors is a solution of complex optimization problem and at the same time combined with the protection of personal data. The problem is resolved at the server level. The driving factors for solving the aforementioned optimization problems are: a) hypergraphs, which can model correlations from a multitude of heterogeneous sources, and b) dynamic models of co-operative filtration (e.g. kalman cooperative filter, dynamic hypergraph factoring) that take into account the time progression of each low-dimension latent component representing it as a multidimensional Brown motion. Connecting tissue between the two driving levers are graphical models and approximate inference (variational inference). At the client level, the development of applications (e.g. iphone/android apps) for mobile phones or tablets that take advantage of anthropocentric interaction in an inherently multilingual environment is required. An open problem in anthropocentric interaction is the identification of main names (e.g. place names), as well as keyword spotting in a noisy multi-language environment. In this project solution to this problem is pursued by the use of deep neural networks. Deep neural networks are also offered for the performance of verbal descriptions of events (e.g. from Wikipedia) in other languages when they are absent. System integration also addresses the minimization of the amount of data to be transferred between client applications and the server, incorporating in client applications e.g. the export of features from videos, images, music clips. At both server and client levels, the challenges faced are significant and timely, since they are research subjects that are of international concern.

Type and scope of work provided

The services provided include the following:
– Requirements analysis – use case scenarios for end software product
– Drafting of Algorithms for analysing touristic preferences and behaviours towards recommending destinations (places, hotels and amenities)
– System integration
– Pilot case with tablet applications for hoteliers
– Testing and debugging of integrated solution
– Dissemination of project results: scientific papers, conferences, networking events, conferences.