Eurocode 2 Review: Decisions and further work

Prepared by Ian Unwin, Food Information Consultancy

in collaboration with
Anders Möller, Food Informatics, Danish Veterinary and Food Administration

Introduction

A report entitled Eurocode 2 Review and modified Core Classification is now available from the Danish Veterinary and Food Administration. A summary of the main findings of the report has been produced to accompany this note on the decisions and further work needed for the development of Eurocode into a widely used and effective approach to the coding of food consumption data and other food-related information. These will be discussed at a COST Action 99 Eurofoods workshop being held in Wageningen on 25 October 1998.

Decisions on the classification

At present the structured codes of the Eurocode 2 classification determine the order in which the classification categories are listed but coders need this order to be logical. Since the present classification does not include vacant codes, insertion of new codes is not possible and updating Eurocode 2 involves changing codes to maintain the logical order of the categories. This loses compatibility with coding that used an earlier version. It is proposed to decide that existing codes should not be changed when the classification is updated.

This requires that the Eurocode documentation system implements a separate mechanism for maintaining the logical order of the classification categories but food coding could still involve the direct assignment of Eurocode 2 codes. However many of the constraints of the present system arise from the need to keep the codes manageable by coders. A decade on from the original work, it can now be assumed that computer support for coding will be the normal working environment, with the selection of an appropriate text term as the preferred approach. Therefore it is proposed to decide that future modifications of Eurocode need not assume that coders will directly use the classification's storage codes.

These decisions would introduce new opportunities for flexibility into the system. Separating management of logical order from the codes would allow alternatives to be defined for the convenience of coders and also as the basis for alternative aggregations of data. Use of codes simply for data storage would remove the constraint limiting the hierarchy to three levels. Indeed since the order and hierarchical relationships of the categories can be handled separately from the codes, the latter need not hold hierarchical information. Although these proposals provide great flexibility, it should be noted that this should be used with care. There should be an agreed standard order and hierarchy, with each main food group having a number of levels appropriate to its requirements. It is proposed that the number of hierarchical levels available in the Eurocode 2 classification should be increased.

Decisions on other parts of the Eurocode system

As well as the Eurocode 2 classification, the Eurocode food coding system includes the food descriptor and recipe recording systems, together with the overall documentation that needs to form a comprehensive user manual. The Eurocode 2 Review suggested that these should be further integrated into the system, which would also allow clearer guidelines and examples to be given to document their use. The present draft version of the User Manual provides a basis for describing and documenting the Eurocode system but it does contain errors, inconsistencies and misleading examples. Also it should be conveniently available to those with an interest in using and developing the Eurocode system and there should be an effective mechanism for the feedback of suggestions, queries and corrections. It is proposed that based on the draft Manual, a revised version of Eurocode 2 should be available on the World Wide Web.

The Review identified various improvements and enhancements to the descriptor system but basic decisions on the recipe recording system are more important. The content of recipe files needs to be specified and a sample set of recipes to be collected. This would provide a prototype for the national recipes envisaged for Eurocode and could use recipes as published in a national food table if an administration publishing a suitable set agrees to this. It is proposed that efforts should be made to find one or more national administrations willing to allow their published recipe details to be made available on the Web.

Further work on Eurocode development

The decisions in section B can be taken in principle at the workshop and those in section C can be implemented if the necessary Web document preparation is undertaken, a task that is seen as essential to widening the interest in Eurocode. As the COST Action 99 LanguaL workshop held in May 1997 decided that other classification systems such as Eurocode 2 could be incorporated into LanguaL facet A, the revised Eurocode document would also serve as the description for its use within LanguaL. However modification of the Eurocode system will involve further work and a list of the tasks considered necessary is included in the report of the Review. The approach taken in the tasks depends on the decisions of the previous two sections and discussion of the requirements for the Eurocode 2 system which will form an item on the workshop agenda.

The first tasks will be to obtain feedback on the core classification as presented in the report and on the overall requirements for the system, for example in respect of matching consumption and composition records. This information will be used to formulate an overall framework integrating use of the classification, descriptor system and recipe records. Clear statements of the policies underlying the classification categories and use of the other parts of the system should form key parts of the user documentation to aid coding and thus these will be incorporated as revisions to the draft Manual. This work would then be used to revise and complete the core classification, using the greater flexibility available if the decision on the extended hierarchy is positive. It would also apply categorisation based on biological origin or product type in accord with policies for each main group.

Work is also required on the descriptor and recipe systems, including their documentation, based on the findings of the report and the further assessment of requirements. The requirements for the management of recipe information relate to the COST Action 99 work on recipe description. Particular concerns are whether the coding of ingredients using Eurocode is sufficiently specific and the formulation of a framework that provides a viable mechanism for the maintenance of recipe files. One particular problem is that variations on a standard recipe, for example through the use of different milks, should not be independent of each other but may generate different codes. As well as managing information on recipe dishes, the recipe system may also be required to hold information on aggregated 'foods', for example consisting of a set of individual items weighted according to their market share.

The final objective should be the design and building of a thesaurus. As well as being a key part of the user documentation, this would form the basis for coding support software allowing coding to be undertaken through the selection of text-based terms, including synonyms. The present draft Manual includes features that can be incorporated and developed into this, in particular the Index of Food Names. The step-wise development of the existing Eurocode resources, together with clear requirements and objectives for enhancing the system, should enable effective coding facilities to be developed and made more widely available within a manageable work plan.

Ian Unwin
3 February 1999