magistrsko delo
Lara Erjavec (Author), Mojca Jevšnik (Reviewer), Andrej Ovca (Mentor), Jelena Ficzko (Co-mentor)

Abstract

Uvod: Sodobni izzivi na področju varnosti živil ter naraščajoča potreba po naprednih rešitvah so spodbudili uporabo umetne inteligence v sistemih za zagotavljanje varnosti hrane. Vključevanje novih sistemov za zagotavljanje varne in kakovostne hrane ter skladnosti s standardi ostaja ključno za ohranjanje zdravja prebivalstva. Poznamo več različnih tehnologij umetne inteligence, ki so že bile uporabljene v živilski industriji: ekspertni sistemi, mehka logika, nevronske mreže, strojno učenje in jezikovni modeli. Poglobljeno se preučujejo prednosti umetne inteligence, kot so hitrost analize podatkov, avtomatizacija odločitvenih procesov in izboljšana natančnost napovedovanja tveganj. Namen: Oceniti potencial uporabe jezikovnih modelov umetne inteligence v kontekstu sistemov za zagotavljanje varnosti živil, s poudarkom na analizi dejavnikov tveganja. Metode dela: Metodološki pristop temelji na teoretičnem pregledu literature, analizi in primerjavi jezikovnih modelov ter študiji primera. Študija primera zajema testiranje izbranega umetnointeligenčnega sistema na dveh teoretičnih in dveh realnih primerih iz živilske industrije po načelih sistema HACCP. Rezultati: Rezultati kažejo, da je umetnointeligenčni sistem najbolj primerljiv z analiziranimi izbranimi primeri za mikrobiološke dejavnike tveganja, tako pri prepoznavanju (48 %), kot pri obvladovanju tveganj s preventivnimi ukrepi (42 %). Sistem je manj primerljiv za kemijske in fizikalne dejavnike tveganja ter alergene. Visoko stopnjo skladnosti z analiziranimi primeri je umetnointeligenčni sistem dosegel pri določanju kritičnih kontrolnih točk, saj je ta bila med 79 % in 92 % glede na izbrane primere. Pri določanju kritičnih mejnih vrednosti, korektivnih ukrepov, vzpostavljanju sistema monitoringa, postopkov dokumentacije in verifikacije je bil umetnointeligenčni sistem v večji meri manj natančen, navedbe so bile bolj splošne in manj prilagojene na specifičen proizvodni proces, so pa večkrat vsebovale »avtomatizacijo« in »digitalizacijo«. Razprava in zaključek: Primerjava rezultatov z obstoječimi zapisi HACCP študij kaže, da umetna inteligenca lahko izboljša analizo dejavnikov tveganj, a popolna avtomatizacija v tem trenutku, brez dodatnih prilagoditev orodja, ni priporočljiva. Pri tem je potrebno poudariti, da je skladnost umetnointeligenčnega sistema temeljila na vsebinski primerjavi s primeri, opisanimi v strokovni literaturi in obstoječi realni HACCP študiji. Strokovna presoja in regulativni nadzor ostajata ključna za zagotavljanje varnosti živil. Nadaljnje raziskave bi tako morale vključevati namensko prilagoditev oz. izboljšanje algoritmov ter redno preverjanje skladnosti z zahtevami zakonodaje ter živilskih standardov.

Keywords

magistrska dela;sanitarno inženirstvo;umetna inteligenca;živilska industrija;varnost hrane;HACCP;

Data

Language: Slovenian
Year of publishing:
Typology: 2.09 - Master's Thesis
Organization: UL ZF - University College of Health Studies
Publisher: [L. Erjavec]
UDC: 614
COBISS: 231550979 Link will open in a new window
Views: 132
Downloads: 48
Average score: 0 (0 votes)
Metadata: JSON JSON-RDF JSON-LD TURTLE N-TRIPLES XML RDFA MICRODATA DC-XML DC-RDF RDF

Other data

Secondary language: English
Secondary title: Possibilities of using artificial intelligence tools in the field of food safety systems
Secondary abstract: Introduction: Modern challenges in food safety and the growing need for advanced solutions have driven the adoption of artificial intelligence in food safety management systems. Implementing new systems to ensure safe, high-quality food and standards compliance remains crucial for public health. Various artificial intelligence technologies have been used in the food industry, including expert systems, fuzzy logic, neural networks, machine learning, sensors, and language models. The benefits of artificial intelligence— such as rapid data analysis, automated decision-making, and improved risk prediction accuracy—are being thoroughly examined. Purpose: To evaluate the potential application of artificial intelligence language models in food safety management systems, focusing on risk factor analysis. Methods: This study is based on a theoretical literature review, analysis and comparison of language models, and a case study. The case study tested a selected artificial intelligence system on two theoretical and two real-world examples from the food industry following HACCP principles. Results: The results show that the artificial intelligence system is most comparable with benchmark cases for microbiological risk factors, both in detection (48 %) and in managing risks through preventive measures (42 %). The system is less comparable for chemical and physical risk factors and allergens. A high degree of compliance was achieved in determining critical control points, ranging from 79 % to 92 % in the selected examples. However, when establishing critical limit values, corrective actions, monitoring systems, documentation procedures, and verification protocols, the artificial intelligence system was considerably less precise; its outputs were more general and less tailored to the specific production process, often mentioning »automation« and »digitization«. Discussion and conclusion: A comparison with existing HACCP study records indicates that while artificial intelligence can enhance risk factor analysis, complete automation at this stage—without further tool adaptations—is not advisable. The consistency of the artificial intelligence system was based on content comparisons with examples from scientific literature and existing HACCP studies. Expert evaluation and regulatory oversight remain crucial for ensuring food safety. Future research should focus on targeted algorithm improvements and regular verification of compliance with legal requirements and food standards.
Secondary keywords: master's theses;sanitary engineering;artificial intelligence;food industry;food safety;HACCP;
Type (COBISS): Master's thesis/paper
Study programme: 0
Thesis comment: Univ. v Ljubljani, Zdravstvena fak., Oddelek za sanitarno inženirstvo
Pages: 73 str., [3] str. pril.
ID: 26171632