Veterinary session

Food chain safety starts by the land and ends at the consumer's table. While the first stages of the food chain are covered by legislation and the activities of operators are controlled by public authorities, the food safety practices of consumers are not covered. Unfortunately, most food-borne illnesses are caused by poor household food hygiene practices. Risk communication is an appropriate tool to improve risk perception and risk-averse behaviour of the population. The scientific and practical basis for risk communication is constantly evolving, as it is necessary to monitor changes in consumers' habits and knowledge, the impact of the information they receive and other factors influencing their behaviour. This research aims to contribute to this by analysing the texts of today's popular online recipe sharing platforms to identify the most common food hygiene mistakes and to propose risk communication measures to mitigate them.

Based on the experience of the literature review, the research focused on recipes using raw chicken meat as a raw material. The text of 500 recipes from the most popular recipe sharing websites, which are open to the general public, were analysed. The food preparation practices collected during the analysis were classified according to the food hygiene recommendations of the National Food Chain Safety Office. A total of 132 poor food hygiene practices were identified, indicating that risky food preparation instructions were present in more than a quarter (26.4%) of the recipes. A further 48 potentially poor food hygiene practices were found in 9.6% of recipes.

Household kitchens are the last line of defence for food chain safety, but the results show that information in online recipes does not prepare consumers for this vital role. Information on the correct preparation of raw chicken should be emphasised in risk communication campaigns aimed at raising food safety awareness among both adults and children. The research has also highlighted the usefulness of developing a web-based text-monitoring robot for machine learning text analysis of this type of web content which could assist in risk management endeavors.