Abstract : Odour characterization from indoor environments is necessary to evaluate perceived indoor air quality or to
apprehend the impact of different sources such as ground and wall coverings. Electronic nose technology seems
interesting face to time -consuming sensory analysis with trained panels. The sensor array used in this study is
based on conducting polymers.
In order to understand what type of information comes from the sensor, the responses to 10 emissions of
indoor frequently used paints where confronted to odour intensity evaluation of these samples by 13 individuals
(olfactory matching method) and to their respective analytical composition. Correspondence Analysis of the
different experimental matrixes (10 paints for 13 subjects, 28 sensors and 27 analytical variables) yielded three
responses spaces. As samples show dependency, the olfactory space gives only three dimensions, the principal
one characterising trigeminal sensation which were confirmed by strong concentrations of aldehydes in alkyd
paints. This dimension is the only one retained in sensor interaction space and the most important in the
analytical space. Paints classification in the 1-dimensional sensor space or in the first factor of analytical space
follows the one found by a specific anosmic individual, whose sole presence increased the number of paints
classes from 2 to 3. This result enhances the need to consider interindividual differences in odour
A four-dimensional space including all data has been defined. The samples that cause trigeminal sensations
were removed. The remaining data were transformed in ranks. Variables classification allows the identification
of five clusters, supposed to correspond to independent ligand-olfactory receptor properties. Classification
similitudes were observed between several sensors and some individuals, indicating similar response
mechanisms. This approach brings some knowledge on the structure of human olfactory receptors and the
number of receptor classes in the long term.
As sensors do not take into account the totality of sensory information, the necessity of sensor technology
hybridization arises as well as the need to integrate new systems more sensitive based on mass spectrometry.
Identical sensor repetition may prove beneficial as a signal convergence tool insuring significant stability and
sensitivity gain. Likewise, the analytical method chosen needs the multiplication of different sampling
techniques and elaborated separative systems coupled to olfactory sniffing in order to hope to identify and to
quantify all of the odorous compounds.
Such evolutions are a necessity before attempting to model olfactory information processing by elaborated
pattern recognition techniques.