Prototyping a decision support system based on semantic web technologies to aid consumers with food sensitivities in their assessment of product safety

摘要

The thesis depicts the information need people with food sensitivities experience in theshopping situation. Diversity within the group is illustrated through five personas. Existingmeasures aimed at helping affected individuals in their search for safe food – ranging fromlabeling requirements and practices to various forms of information systems – areexemplified. Shortcomings of existing solutions are discussed from an information scienceperspective. An alternative approach based on Semantic Web technologies and Linked datais proposed, underpinning a decision support system. The Boolean interpretation of foodsafety is rejected in favor of a division that accounts for the need for human assessment ofuncertain cases. SPARQL and automatic inference, relying on a dataset holding facts aboutallergen occurrence in various ingredients, is used to automatically classify products as safe,uncertain or unsafe for individual users. Proof of concept for the proposed approach isprovided through a prototype web application. The automated classification is used tocommunicate the safety of each product, using familiar traffic light colors. Tailored decisionsupport is provided on-demand, emphasizing information that is likely to impact the usersassessment of uncertain products while limiting “information overload”. The developmentprocess behind the ontology and web application is discussed in detail, followed by adiscussion about how to establish the required data sets.