Chemometric Modeling of Coffee Sensory Notes through Their Chemical Signatures: Potential and Limits in Defining an Analytical Tool for Quality Control
| Autor: | Carlo Bicchi, Patrizia Rubiolo, Davide Bressanello, Erica Liberto, Manuela Rosanna Ruosi, Chiara Cordero, Barbara Sgorbini, Gloria Pellegrino |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Quality Control
Computer science media_common.quotation_subject HS-SPME-GC-MS Sensory system Sensory profile Machine learning computer.software_genre Coffee 01 natural sciences Gas Chromatography-Mass Spectrometry Chemometrics chemometrics coffee aroma sensory note fingerprints Chemistry (all) Agricultural and Biological Sciences (all) 0404 agricultural biotechnology Discriminative model Food Quality Humans Quality (business) Solid Phase Microextraction Aroma media_common Principal Component Analysis Volatile Organic Compounds biology business.industry 010401 analytical chemistry Reproducibility of Results 04 agricultural and veterinary sciences General Chemistry biology.organism_classification 040401 food science 0104 chemical sciences Models Chemical Taste Odorants Artificial intelligence Wine tasting General Agricultural and Biological Sciences business computer Predictive modelling |
| Zdroj: | Journal of Agricultural and Food Chemistry. 66:7096-7109 |
| ISSN: | 1520-5118 0021-8561 |
| Popis: | Aroma is a primary hedonic aspect of a good coffee. Coffee aroma quality is generally defined by cup tasting, which however is time-consuming in terms of panel training and alignment and too subjective. It is challenging to define a relationship between chemical profile and aroma sensory impact, but it might provide an objective evaluation of industrial products. This study aimed to define the chemical signature of coffee sensory notes, to develop prediction models based on analytical measurements for use at the control level. In particular, the sensory profile was linked with the chemical composition defined by HS-SPME-GC-MS, using a chemometric-driven approach. The strategy was found to be discriminative and informative, identifying aroma compounds characteristic of the selected sensory notes. The predictive ability in defining the sensory scores of each aroma note was used as a validation tool for the chemical signatures characterized. The most reliable models were those obtained for woody, bitter, and acidic properties, whose selected volatiles reliably represented the sensory note fingerprints. Prediction models could be exploited in quality control, but compromises must be determined if they are to become complementary to panel tasting. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|