Food Informatics Unilevers BitterBase Case

1
Food InformaticsUnilevers BitterBase Case

• Description and Decomposition

Hilbert Bruins Slot
2
BitterBase

• Knowledge repository to understand the principles
  of bitterness and to provide ways to prevent
  bitterness
• A database of bitter ingredients, bitter maskers
  and debittering routes
• Available information on bitter ingredients has
  been collected (gt3000 papers, patents 300/year)
• 3000 entries molecules, mixtures
• Molecular module in preparation that categorises
  molecules on the basis of structure, enabling
  bitterness prediction and suggestions for masking

3
Bitterness a subjective experience

• Functional ingredient might be bitter tasting
• Processing (heating, cooling etc) might cause
  bitterness
• Combined sensations (bitter and sweet)
• Bitterness as an acquired taste (psychology,
  culture)
• Super- and Non- tasters
• The difficulty complexity of bitterness
• 26 different receptors
• Different classes of bitter ingredients require
  different masking solutions
• Complex processes of activation/inhibition and
  masking

4
BitterBase use

• To find
• information on bitter ingredients
• solutions for masking or debittering
• alternative ingredients without bitter sensation
• masking or debittering solutions based on
  molecules of the same class
• predicting bitterness of ingredients (future)
• predicting the masking of bitter ingredients
  (future)

5
Different categories in BitterBase

• Display Name
• Name
• Taste
• Masking
• Receptor
• Source
• Use
• Hazard
• Threshold
• Miscellaneous
• Structure_type
• Cas
• Structure

Activity analysis with AID "BitterBase consists
of a set of tables with information about
molecules, thresholds, the receptors they
affect, text fragments of reports etc." Aim of
AID in this context would be to create a much
richer, more flexible, adaptive environment,
that would give better support to researchers.
6
High level view of main activities
7
Cumbersome, highly manual information
gathering and codification
8
Curator is responsible for quality of database
content
9
Simple boolean queries resulting in list of
ingredients
10
(No Transcript)
11
Overview of main user interactions of a future,
enriched (ontologies, causal models, etc.)
Bitterbase
12
Supporting causal interactions Learning
new hypotheses about ingredients / receptors /
interactions Identify hidden patterns in
knowledge base Updates semi-automatically with
external sources Allowing for answering complex
questions What are the trends in masking /
encapsulation Which authors are experts for
which bitter compounds Which masking strategies
are used for which bitter molecules Can we
cluster bitter molecules on the basis of text
13
Some processes and routines to be designed
14
Unilevers BitterBase Case