14B-5

N. TIPSRISUKOND, I. U. Gruen, J. Tan, M. A. Beierschmitt, J. Ji, L. Tu, F. N. S. Weerasinghe, and L. N. Fernando. Department of Food Science, University of Missouri, 122 Eckles Hall, Columbia, MO 65211-5200

Wine has long been known as a healthful beverage and is relished as a food and/or food adjunct. Consumers have specific expectations about the type of wine that is to be consumed with certain foods, presumably due to the flavor profile of the wine. However, very few studies have addressed the question of the relationship between a particular wine aroma and a specific food.

The objectives of this study were to determine the headspace volatiles of four different wines using SPME, to distinguish the wines using headspace volatiles and to correlate the chemical data with both sensory descriptive and hedonic results using fuzzy sets and neural networks.

Solid-phase microextraction, mass spectrometry, multivariate analysis (SPME-MS-MVA), descriptive analysis (DA), hedonic test, fuzzy sets and neural networks were utilized for the study of volatile compounds in wines, including Cabernet Sauvignon (CS), Pinot Noir (PN), Chardonnay (CD) and Sauvignon Blanc (SB).

SPME allowed fast, inexpensive, relatively easy operation and its reproducibility was acceptable. Twenty-nine wine volatiles were tentatively identified. SPME-MS when combined with MVA revealed that 1-dodecanol, furfural and 3-propylcyclopentene were responsible for red wine characteristics and were correlated with tough and tannic attributes. Ethyl esters were the primary volatile compounds detected in CD, while p-cymene and octanoic acid were important volatiles in SB. A highly positive correlation between the sour attribute and ethyl esters was discovered. Fuzzy sets and neural networks could be utilized to determine the relationship between DA and hedonic results, despite the small number of samples.

Fuzzy sets and neural networks provide a very useful technique to correlate chemical and sensory data.