*2.1. EuroFIR AISBL's Approaches*

Many international projects and research networks have tried to standardize methods for collection, management, and publication of food data. Efforts in the development of procedures to define and establish standardized collections of food composition data, specifically nutrient content, have also been carried out (e.g., description, selection, preparation, references, analytical or computational approach, compilation) [5,31,32]. EuroFIR AISBL, along with national compilers, have put considerable effort, now recognized globally, in establishing standardized and harmonized food datasets to assure the quality of both compilation processes and their presentation [1,2,5,23–27,33–44].

#### 2.1.1. Quality Management System and Standard Operating Procedures

To ensure the conformity (interoperability) of FCDBs, datasets must demonstrate transparency in aggregation, validation, and compilation based on standardized documentation and quality evaluation. EuroFIR AISBL has established a quality-data management system and harmonized and standardized processes.

EuroFIR AISBL's standard operating procedures (SOPs) are identified in various publications such as: (i) documentation of foods, nutrients, and background information (i.e., metadata); (ii) attribution of quality indices to original data; (iii) coding of original data before data entry; (iv) quality check on original data coding and data entry; (v) physical storage of original data; (vi) selection of original data for aggregation; (vii) selection and application of algorithms to produce aggregated and compiled datasets; (viii) validation of aggregated and compiled datasets; and (ix) selection of aggregated and compiled datasets for dissemination as a published database or tables as well as guidelines for quality data evaluation [23,45,46].

Documentation of information concerning foods, components, values, and references is essential in maintaining a FCDB. One working groups developed guidelines for default value documentation of aggregated/compiled values using the EuroFIR AISBL's standard and thesauri. Options for aggregation/compilation in the FoodCASE data management system were taken as the starting point [23].

#### 2.1.2. Food Description and Classification Systems

There is a consensus on the importance of nomenclature, (food) descriptions, and classification of foods. In this context, and with a view to the exchange of data, design, and development of a database primarily requires exact identification of a food. FoodEx2 is a standardized food classification and description system developed by EFSA, and supported by FAO INFOODS, to describe characteristics of foods and food supplements in exposure assessment studies. This system comprises flexible combinations of classifications and descriptions based on a hierarchical system for food safety-related domains (i.e., food consumption, contaminants, pesticide residues, veterinary drug residues, zoonoses– biological and microbiological aspects, botanicals, and food composition) [47–51].

LanguaLTM or "Langua aLimentaria" or "Language of food" (LanguaLTM) is generally recognized as a method for describing foods, facilitating the capture and exchange of food data. More specifically, LanguaLTM has a controlled vocabulary for systematic food descriptions that can be used with thesauri for faceted classification [52]. As described by Møller and Ireland [53], any food (or food product) can be described systematically using a combination of characteristics. In turn, these characteristics can be categorized and coded for computer processing, and resulting viewpoint/characteristic codes can be used to retrieve data about foods from similarly coded external sources. Each food is described using a set of standard, controlled terms taken from facets characteristic of the nutritional and/or quality aspects of a food, such as: food source, i.e., ANIMAL USED AS FOOD SOURCE [B1297], PLANT USED AS FOOD SOURCE [B1347]; cooking, i.e., TOASTED [G0010], BOILED [G0014], STEAMED WITH PRESSURE [G0022], DEEP-FRIED [G0029]; preservation methods, i.e., PASTEURIZED BY IRRADIATION [J0119], PRESERVED BY FREEZING [J0136], PRESERVED BY STORAGE IN CONTROLLED ATMOSPHERE [J0176]; and treatment applied, i.e., BLEACHED [H0197], PUFFED [H0268], EXTRUDED [H0352]. Several applications of simple foods, food preparations, recipes, food supplements, and agro-food wastes have been carried out [54–57].

LanguaLTM was developed principally to support data exchange, whereas FoodEx2 was developed as a food classification and description system for exposure and risk assessment studies, i.e., exposure to contaminants. LanguaLTM codes are assigned following a facet scheme set in advance, which defines and describes foods (i.e., source, origin, physical state, heat treatment, cooking method, treatment, preservation, packaging, etc.), and this scheme must be applied and maintained for all food items. FoodEx2 coding aggregates food products according to need without following a pre-agreed scheme. For instance, POACHED EGGS are codified by FoodEX2, revision 2 [47–51] using a single base term [A032D], whereas LanguaLTM codifies them using terms string: 02 EGG AND EGG PRODUCTS (EUROCODE2) [A0725], HEN [B1713], WHOLE EGG WITHOUT SHELL [C0225], WHOLE, SHAPE ACHIEVED BY FORMING [E0147], FULLY HEAT-TREATED [F0014], SIMMERED, POACHED OR STEWED [G0020], HUMAN CONSUMER, NO AGE SPECIFICATION [P0024]. Recently, examples of applications using both systems on food preparations and recipes were given by Durazzo et al. [54]. FoodEx2 uses implicit descriptors to reduce code length, whereas LanguaLTM descriptors are fully explicit and structured. Both LanguaLTM and FoodEx2 are updated regularly based on feedback from users. User training courses are run for both LanguaLTM and FoodEx2.

LanguaLTM and FoodEx2 are the main food description and classification systems, and both are well developed, widely used, and recognized at European and International levels [42]. Their use also represents the likely direction of future work, specifically the automation of matching, mapping, and data quality checking. Consequently, maintenance and updating of both systems must be carried out regularly through exchanges between users and developers, considering evolution of the food market and new food classification needs in different applications. Subsequently, the correct application of classification and description systems relies on standard operating procedures (SOPs), regular updates, and multi-disciplinary cooperation [42].

These schemes are, however, not the only coding approaches, and their use can be supplemented with other systems such as ontologies. FoodOn is an open source, harmonized, and comprehensive food ontology that supports global food traceability, quality control, and data integration [58,59]. It is composed of term hierarchy facets that cover basic raw food source ingredients; process terms for packaging, cooking, and preservation; and an upper-level variety of product type [58,59]. For nutrient composition, and likely also bioactives and botanicals, however, EuroFIR AISBL recommends ongoing activities use of LanguaLTM and FoodEx2 [42].
