template-browser-not-supported

Noticias

Developing an online restaurant recommendation system that predicts which picture an internet user will take

The research, carried out by the University of Oviedo, assumes that the images are what can best convince them

Research by the University of Oviedo has developed a system of online restaurant recommendations similar to those recommended on a digital platform, or that show the purchases that could be made by a subbject that we could examine in an online store. To do this, the research staff designed and implemented an algorithm that makes individual predictions of the photos users might take if they went to a restaurant - bearing in mind the user's uploaded photograph history. The images are what is best able to convince them of the establishment's qualities, given that whoever takes the photos does so to reinforce or justify their opinions.

The article, which was published in Information Sciences (one of the best scientific journals in its field) to illustrate the project, used data collected (from 6 different sized cities) from a well-known website containing restaurant reviews with photos. The system would say to potential visitors: "Go to this restaurant that you'll like, and this is the picture you might take."

The research was carried out at the University of Oviedo's Artificial Intelligence Centre, led by Professor Antonio Bahamonde, together with Beatriz Remeseiro, Oscar Luaces, Pablo Pérez-Núñez and Jorge Díez.

The progress made does not end with making predictions of attractive images, but also opens the door to estimating how they will be distributed. This means a company will be able to see the products and features highlighted by customers.