Predicting organoleptic scores of sub-ppm flavour notesPart 2.† Computational analysis and results

Abstract

In Part 1 of this paper (T. C. Pearce and J. W. Gardner, Analyst, 1998, 123, 2047) we describe a novel method for predicting the organoleptic scores of complex odours using an array of non-specific chemosensors. The application of this method to characterising beer flavour is demonstrated here by way of predicting a single organoleptic score as defined under the joint EBC/ASBC/MBAA international flavour wheel for beer. An experimental study was designed to test the accuracy of the odour mapping technique for this prediction of organoleptic scores of added reference compounds within a chemically complex lager beer background. Using the flow injection analyser (FIA) system comprising 24 conducting polymer sensors, also described in Part 1, sampling was conducted on spiked lager beers. A dimethyl sulfide spike was added at the 20–80 ppb v/v level to simulate a range of organoleptic scores (0–5.5 out of 10) for flavour note no. 0730—“cooked vegetable”. A certain amount of sensor drift was observed over the 12 d testing period which is shown to account for significant variance in the data-set as a whole. The effect of the sensor drift was reduced by applying a linear drift model, which may be generally applied when the between-class variance due to the difference in odours is small when compared with the within-class variance due to the drift, which increases approximately linearly over time. Careful use of this drift compensation model, coupled with judicious selection of pre-processing and pattern recognition techniques, maximised the between-class variance and so improved the overall classification performance of the system. After applying detailed exploratory data analysis, statistical, and neural classifier techniques, the organoleptic score was predicted with an accuracy of ±1.4 (out of 10) and 95% confidence. Our results show that it is possible to generate subjectively defined organoleptic flavour information using multi-sensor arrays and associated data-processing that is comparable in accuracy to sensory and GC-based techniques.

References

1. T. C. Pearce and J. W. Gardner, Analyst, 1998, 123, 2047.
2. ASBC Technical Committee and Editorial Committee, Methods of Analysis of the American Society of Brewing Chemists, ASBC, St. Paul, USA, 1976.
3. H. F. Kaiser, EPM, 1960, 20, 141.
4. M. S. Bartlett, British Journal of Psychology, Statistical Section, 1950, 3, 77.
5. W. J. Krzanowski, Biometrics, 1987, 43, 575.
6. C. DiNatale, F. A. M. Davide and A. D'Amico, Sens. Actuators B, 1995, 27, 237.
7. M. Holmberg, F. Winquist, I. Lundström, F. Davide, C. DiNatale and A. D'Amico, Sens. Actuators, 1996, 36, 528.
8. V. Barnett, Journal of the Royal Statistical Society, Series A, 1976, 139, 318.
9. S. S. Stevens, Psychophysics, McGraw-Hill, New York, 1975.