Consumer Language‐Based Sensory Descriptor Extraction and Volatile‐Sensory Associations in Aged Light Flavor Baijiu

Sensory evaluation of aged light-flavor baijiu (LFB) traditionally relies on trained panels and expert-generated lexicons; however, standardized descriptors may not fully reflect the terminology used by consumers during actual consumption. This study aimed to extract consumer-derived sensory descriptors from online reviews and integrate them with trained-panel evaluation and instrumental flavor analysis to characterize volatile compounds evolution during aging of LFB. A total of 9596 online review text segments were manually mined to extract flavor descriptors, which were subsequently used as sensory attributes for quantitative descriptive analysis (QDA). Electronic nose and headspace solid-phase microextraction Arrow gas chromatography-mass spectrometry (HS-SPME-Arrow-GC-MS) were used to characterize volatile profiles, and redundancy analysis (RDA) was applied to investigate associations between volatile compounds and sensory attributes. A total of 103 volatile compounds were identified, including 40 esters, 19 alcohols, 8 aldehydes, 7 acids, 3 ketones, 6 phenols, 12 terpenes and terpenoids, and 8 other compounds. Esters were the most abundant volatile fraction, increasing from 7359.99 to 13,611.60 µg/L during aging, whereas acids and phenols showed an increasing-then-decreasing trend. QDA showed that aged samples had higher floral, fruity, mellow, and lasting-aftertaste scores. RDA explained 49.85% of the total variance and revealed significant associations between consumer-derived descriptors and key volatile compounds. These findings provide a consumer-oriented framework linking online consumer language with chemical markers, supporting quality grading, sensory evaluation, and optimization of aging processes for LFB.