Points-of-Interest Recommendation Algorithms for a COVID-19 Restrictions Scenario in the Catering Industry

Abstract

The COVID-19 pandemic has affected many areas of day-to-day life, including tourism and restaurants. Many countries imposed restrictions on restaurants during the COVID19 period. Many restaurants closed, and others switched to delivery and take-out services. These restrictions affect both the catering system as a whole and smart catering systems, such as recommender systems and user experience aggregators. The main purpose of the article is to assess the impact of COVID-19 on these digital components in different countries, depending on the COVID-19 strategy. In particular, the author’s contribution is as follows: (1) assessing the stability of recommendation algorithms depending on the country’s COVID-19 elimination strategy, (2) identifying factors associated with changes in user behavior during the COVID-19 pandemic, (3) using these factors to improve the recommendation system, (4) answering the counter-question of whether the actual quarantine compliance can be determined using these data. As a result of the experiments, we have identified a change in the accuracy of recommendation algorithms both during and after the lockdown. We also obtained factors for changing user behavior and made assumptions about quarantine compliance in various countries using user experience data. The proposed contextual method has shown increased efficiency during the COVID-19 period.