From these viepoints; this paper examines a new product approach based on various emotional data of human being; such as on kansei engineering; to improve the traditional processing on intuition and previous experiences for product development in the food industry. This approach uses a neural network; which is composed of data trains from the actual products ranked for large sales or poor sales. At first; each input data for neural network is transformed into a numerical value from product data which mean taste; appearance; slicing and the other 6 items based on market research in advance (Endo; 2005; 2005; 2006). The output data are also classified into 3 items corresponding to sales figures based on the total sales of all products. Next; the neural network is trained using the data group for learning. Thereafter; the data group for verification is also evaluated using the trained neutral network; and the probability of appropriate evaluation is examined. Each group has 150 product data in this experiment. In the experiment; all previous product data are randomly classified into the data group for learning or for verification respectively. And; the system reliance is examined on the extent of coincidence between the results yielded by the system and the current impressions of the goods according to 50 subject membersâ€™ market assessment. The statistical test methods (Asai; 2004); such as Wilcoxon test; show there is little difference in the results between sales rank obtained by using the system and goods image by user impression. These results mean that the higher the sales ranks to products assigned by the system; the better the user impressions to goods on the purchase of pickles. As the results; this system is useful for developing and evaluating tool of goods for large sales trends. Based on these results; This paper says that our prototype system is available to the new support system for product development; and it is expected to be applied to product development toward a new direction in the food industry in the future.
Keywords: Kansei engineering; Subjective data; Neural network; Wilcoxon test; Product development; Pickles