• Reduce text

    Reduce text
  • Restore text size

    Restore text size
  • Increase the text

    Increase the text
  • Print

Rangées de graines.. © INRA, Elena Schweitzer © Fotolia

Our results

  1. Introduction
  2. Research and Innovation 2018 - For Food and Biobased Products
  3. Dry-cured ham: a process simulator can now define routes of manufacture that yield lower-salt products
  4. Does organically-farmed meat contain fewer chemical contaminants?
  5. The way in which proteins aggregate when heated may change their sensitising potency
  6. Enhancing the viability of spray-dried probiotic bacteria by stimulating their stress tolerance
  7. Human milk digestion in the preterm infant: impact of technological treatments
  8. Research & Innovation 2017 - For Food and Biobased Products
  9. To stick or not to stick? Pulling pili sheds new light on biofilm formation
  10. When biopolymers selfassemble: a balance between energy and entropy.
  11. Mimicking the gastrointestinal digestion in a lab-on-a-chip:the microdigester
  12. How a milk droplet becomes a powder grain
  13. Research & Innovation 2016 - For Food and Bioproducts
  14. A new process for the biorefining of plants
  15. Under the UV light : the bacterial membrane
  16. Reverse engineering or how to rebuild ... bread!
  17. Green Chemistry: a step towards lipid production in yeast
  18. Individually designed neo-enzymes for antibacterial vaccines
  19. Multi-scale mechanical modelling: from the nanometric scale to the macroscopic properties of bread crumb
  20. Minimill: 500 g to assess the milling value of soft wheats
  21. Microbial production of lipids for energy or chemical purposes
  22. The discrete role of ferulic acid in the assembly of lignified cell wall
  23. Eco-design of composites made from wood co-products
  24. Analysis of volatile compounds enables the authentication of a poultry production system
  25. Nanoparticles as capping agents for biopolymers microscopy
  26. Pasteurisation, UHT, microfiltration...All the processes don't affect the nutritional quality of milk in the same way
  27. Integration of expert knowledge applied to cheese ripening
  28. Controlling cheese mass loss during ripening
  29. The shape memory of starch
  30. Research & Innovation 2015 - For Food & Biobased Products
  31. Behaviour of casein micelles during milk filtering operations
  32. Overaccumulation of lipids by the yeast S. cerevisiae for the production of biokerosine
  33. Sequential ventilation in cheese ripening rooms: 50% electrical energy savings
  34. An innovative process to extract bioactive compounds from wheat
  35. Diffusion weighted MRI: a generic tool for the microimaging of lipids in food matrices
  36. Characterization of a major gene of anthocyanin biosynthesis in grape berry
  37. New enzyme activity detectors made from semi-reflective biopolymer nanolayers
  38. Improving our knowledge about the structure of the casein micelle
  39. Heating milk seems to favour the development of allergy in infants
  40. Fun with Shape
  41. Using volatile metabolites in meat products to detect livestock contamination by environmental micropollutants
  42. SensinMouth, when taste makes sense
  43. A decision support system for the fresh fruit and vegetable chain based on a knowledge engineering approach
  44. SOLEIL casts light on the 3D structure of proteins responsible for the stabilisation of storage lipids in oilseed plants
  45. A close-up view of the multi-scale protein assembly process
  46. Controlling the drying of infant dairy products by taking water-constituent interactions into account
  47. Polysccharide nanocrystals to stabilise pickering emulsions
  48. Discovery of new degradative enzymes of plant polysaccharides in the human intestinal microbiome
  49. A durum wheat flour adapted for the production of traditional baguettes
  50. Virtual modelling to guide the construction of « tailored-made » enzymes
  51. How far can we reduce the salt content of cooked meat products?
  52. Diffusion of organic substances in polymer materials: beyond existing scaling laws
  53. Smart Foams : various ways to destroy foams on demand !
  54. Dates, rich in tannins and yet neither bitter nor astringent
  55. Sodium content reduction in food
  56. Research & Innovation 2014

Integration of expert knowledge applied to cheese ripening

The aim of this study was to reconstruct the ripening dynamics of a Camembert-type cheese on the basis of overall established knowledge. Within this body of knowledge, expert know-how was considered. This know-how is difficult to collect. This is known as the “paradox of expertise”, a well-known phenomenon in artificial intelligence. At the same time, methods were developed in cognitive psychology to elicit expert knowledge. An application of these methods was successfully developed within the framework of our application. At the same time, on the basis of

1) this expert know-how

2) hypotheses established by scientists to represent physical laws and reaction kinetics during ripening

3) available databases, a model representing the complex reaction dynamics during ripening was developed and validated with over 80% positive responses on a test set

Moulage à la louche de camemberts.. © INRA, FROC Jean

Cheese ripening : a balance between human expertise and biological transformations

Fermentation processes and, more generally, biological transformations for which enzymes and microorganisms are responsible, are very often managed in a relatively empirical way, without any other control than that of environmental conditions.
A great number of technical and scientific impediments exist and explain the reasons for this situation.  Among them, we can include the well-known shortfalls of reliable means of measurement and the excessive cost required to have access to relevant biological scales, the complexity of biological transformation agents and the effect of numerous outside factors on their development and/or their activity, the inadequacy of explanatory models of the phenomena at work and the disparity and even incompleteness of available data.

Within this context, our field of interest deals with the integration of knowledge within the framework of a complex system.  The biological or food-related transformation is assimilated to a complex system in which a network of interactions linking variables involved at different scales is established.  The objective is therefore to control the overall behaviour of the process studied (unit operation or series of unit operations) by reducing the uncertainty that characterises it.  To do this, our research is oriented towards the modelling of a process and its dynamics by integrating the different pieces of the knowledge puzzle (knowledge objects established at different scales and in different formats).

Collecting cheesmakers' expertise and modeling process dynamics

A method for collecting and analysing human expertise based on approaches developed in cognitive psychology and sensory analysis was developed and applied to Camembert ripening.  This was done in conjunction with expert cheesemakers and made it possible to collect and formalise knowledge about process dynamics.  Four ripening stages were defined, as well as sensory indicators that made it possible to identify them.
To validate the knowledge collected, ripening experiments were carried out in a pilot ripening room according to an experimental design that used different ripening temperatures and relative humidities.  The microbiological, physico-chemical and biochemical kinetics of cheeses in the process of ripening were analysed.

Using a logistics model, we established a correlation of 88% between the microbiological, physico-chemical and biochemical data, and the sensory phases measured according to expert knowledge.
On the basis of this data collection and at the same time that this approach was applied, a model representing the reaction kinetics taking place within a Camembert-type cheese in the process of ripening was established (presented in "Controlling cheese mass loss during ripening").
This model, represented in the form of a dynamic Bayesian network, is based on the integration of expert knowledge and physical laws where a part of the parameters are fitted to existing experimental data. It proved itself to be representative of over 80% of the kinetics tested.  

Towards the application on other traditionnal food processes

Concerning the collection and analysis of human expertise, it will be necessary to validate this approach on other types of traditional food processes for which the expertise of operators/craftsmen is at the core of product quality.  If its genericity is confirmed, it could be used, on the one hand, to transmit expert know-how within companies and, on the other hand, to support decision-making systems that use different sources of knowledge available about the given object.
As for research oriented towards the modelling of the dynamics of a food system, questions of the genericity and adaptability of the mathematical tools developed are currently being studied.  They involve different approaches such as the use of mechanistic models in the application considered here, the one describing mass loss of cheeses in the process of ripening, as well as other tools such as dynamic Bayesian networks and neuronal methods in general.
Our aim is therefore to establish the most global (integrated) description possible of process dynamics.

A project funded by the National Agency for Research, in partnership with

ACTILAIT (for cheese ripening experiment) : Jean-René Kerjean.
CREA (Ecole Polytechnique), ISC PIF (Institut des Systèmes complexes de Paris Ile de de France) Paris  : Paul Bourgine.
INRIA Saclay Team APIS Saclay  : Evelyne Lutton.
LIP6 Team DESIR Paris : Pierre-Henri Wuillemin
INRA, UMR GMPA Grignon : Nathalie Perrot

See also

  • See the integrating knwoledge programme
  • Barrière, O., Lutton, E., Baudrit, C., Sicard, M., Pinaud B., Perrot, N. (2008). Modeling Human Expertise on a Cheese Ripening Industrial Process Using GP. Lecture Notes in Computer Science, Parallel Problem Solving from Nature – PPSN X , Springer Berlin/Heidelberg (Eds). Vol 5199, pp. 859-868.
  • Baudrit, C., Wuillemin, P.H., Sicard M., Perrot, N. (2008). A Dynamic Bayesian Network to Represent a Ripening Process of a Soft Mould Cheese. Lecture Notes in Computer Science, Knowledge-Based Intelligent Information and Engineering Systems. Springer Berlin/Heidelberg (Eds). Vol 5178, pp. 265-272.
  • Sicard M., Guillon M. (2008). Comment pérenniser les savoir-faire ? Process-alimentaire, Ed. Dubois-Baudry, p.66, sept. 2008.
  • Sicard M., Perrot N., Leclercq-Perlat M.-N., Corrieu G.,(2008). Towards integration of expert skills and instrumental measurements for monitoring cheese throughout the ripening process: Application to Camembert-type cheeses. International Dairy Journal. (en cours de soumission).