Abstract: In this study, consumer evaluation data on the key sensory attributes of stewed pork with brown sauce from 8 different provincial-level regions were collected by web crawler technology for flavor profiling. A vocabulary of sensory descriptors was established. Multivariate statistical methods such as big data analysis and correlation analysis were used to explore the consumer preferences, the correlation and regional differences of the sensory attributes from the multidimensional perspectives. The results showed that in the past two years, consumers preferred stewed pork with brown sauce which was soft and could be eaten without chewing, was agreeably fatty, and tasted fresh and slightly sweet with a good aroma. There were considerable regional differences in the main sensory attribute, which was determined as agreeably fatty mouthfeel for all 8 regions. Softness and stickiness were the scales for measuring the fatty mouthfeel and were significantly affected by sweetness, pungency and saltiness (P < 0.05). The correlation coefficient between softness and sweetness was 0.707, and those between stickiness and pungency and saltiness were 0.815 and ?0.654, respectively, suggesting that consumers preferred a product with a sweet taste and soft texture or with a pungent and properly salty taste and sticky texture.

Key words: stewed pork with brown sauce, web crawler, big data, descriptors vocabulary, sensory attributes