Wine Experts Research Articles

Using Neural Network Models for Wine Review Classification

AbstractWines are usually evaluated by wine experts and enthusiasts who give numeric ratings as well as text reviews. While most wine classification studies have been based on conventional statistical models using numeric variables, there has been very limited work on implementing neural network models using wine reviews. In this paper, we apply neural network models (CNN, BiLSTM, and BERT) to extract useful information from wine reviews and classify wines according to different rating classes. Using a large collection of wine reviews from Wine Spectator, the study shows that BERT, a neural network framework recently developed by Google, has the best performance. In the two-class classification (90–100 and 80–89), BERT achieves an accuracy of 89.12%, followed by BiLSTM (88.69%) and CNN (88.02%). In the four-class classification (95–100, 90–94, 85–89, and 80–84), BERT yields an 81.57% accuracy, while the other two produce an 80% accuracy. The neural network models in the paper are independent of domain knowledge and thus can be easily extended to other kinds of text analysis. Expanding the limited work on wine text review classification studies, these models are up-to-date and provide valuable additions to wine data analysis. (JEL Classifications: C45, C88, D83)

Effect of Low-Frequency Ultrasonication on Red Wine Astringency

ABSTRACT Wine tastes better with age because of a complex chemical reaction among sugars, acids, and phenolic compounds. This study investigates if applying ultrasonic waves to wine would significantly reduce its perceived astringency levels. Ultrasonic (US) waves were applied to samples of a young Cabernet Sauvignon using a 24 kHz ultrasonic processor and sonotrode probe by varying time,, and amplitude at three levels. To objectively assess the relationship between the US waves and astringency, physical and chemical analysis was carried out that confirmed the underlying assumptions. While the pH of the treated samples decreased slightly, there was no change in color (Hunter CIE Color L*a*b*). Total Phenolic Content (TPC), Total Anthocyanin Content (TAC), and Tannin Concentration (TC) were significantly different (p ≤ .05) among all samples, indicating the impact of sonication on astringency causing components. Additionally, Fourier Transform Near Infrared (FT-NIR) spectroscopy confirmed that there were notable changes in the spectra, attributed to tannins, of wine after the application of sonication in comparison to untreated wine samples. The perception of astringent flavor was evaluated by (i) expert wine tasters and (ii) untrained panelists (n = 60) who were able to clearly distinguish between treated and untreated samples (p ≤ .05) and preferred sonicated samples (180 sec, 100% amplitude) over the control samples, supporting the hypothesis that sonication reduces the astringency of red wine.

Expensive and Cheap Wine Words Revisited

AbstractPrevious work has quantitatively analyzed expert wine descriptions to identify some so-called “expensive” and “cheap” words that are indicative of a wine's price. This paper revisits that work. In particular, I examine whether words previously identified as expensive and cheap ones are still indicative of a wine's price when using the same methods and a different, larger dataset. My findings mostly confirm previous conclusions, although many directions for further research into expensive and cheap wine words remain open. (JEL Classifications: D83, L66)

Do you want a description with that wine? The role of wine mental imagery in consumer's desire to drink using the revised Vividness of Wine Imagery Questionnaire (VWIQ‐II)

AbstractThe influence of general wine imagery vividness on consumers' reported desire to drink was investigated. In Study 1, the Vividness of Wine Imagery Questionnaire (VWIQ) was revised and validated so that it included the dimensions of sight, smell, flavor, and mouthfeel. Mouthfeel is an important factor in wine appreciation, both for consumers and wine experts. In Study 2, we demonstrated the usage of VWIQ in a consumer context: participants were asked to indicate their desire to drink a range of wines that differed in familiarity, with half the participants also receiving a multisensory description of the wine in addition to information regarding the wine's geographical origin and grape variety. Without a description, consumers with higher imagery vividness reported higher desire to drink compared with consumers with lower imagery vividness. However, with a description, the desire to drink from the lower imagery vividness group increased, matching the higher imagery vividness group.Practical applicationThe ability to imagine helps people to plan for the future. In effect, imagery ability can influence how consumers make purchase decisions. Sensory descriptions thus seemed to override differences in people's ability to imagine a wine. In summary, this research demonstrates the value of VWIQ as a tool to tailor advertisements and wine descriptions to specific groups of consumers.

Sensory Relevance of Strecker Aldehydes in Wines. Preliminary Studies of Its Removal with Different Type of Resins.

The orthonasal quality of two synthetic contexts of wine (young wine and oaked wine) spiked with six different levels of the Strecker aldehydes (isobutyraldehyde, 2-methylbutanal, 3-methylbutanal, methional and phenylacetaldehyde) was evaluated by a panel of wine experts. The aldehyde levels simulated the concentrations present in wines protected from oxidation during production and storage and after severe oxidation. Significant quality detriments were observed at concentrations of 13 µg/L of methional, 49 µg/L of phenylacetaldehyde, 17 µg/L of isobutyraldehyde, 12 µg/L of 2-methylbutanal and 24 µg/L of 3-methylbutanal. The presence of these levels of aldehyde concentrations induced the reduction of fruitiness in young wines and of woody notes in oaked wines as well as the appearance of the typical attributes that define wine oxidation. More than 75% of recently opened commercial wines contain total levels of Strecker aldehydes higher than those, however their effect is not always noticeable as they are forming inodorous adducts with SO2. Nevertheless, this content is a potential risk for the shelf life of the wine, as once SO2 is depleted, these aldehydes could release back into their odour-active forms. Thus, in order to reduce the presence of Strecker aldehydes, eight different resins were studied (two scavengers, four mixed-mode anion exchange and two pure anion exchange) in white wine at two levels of SO2. After 24-h contact, the mixed mode Strata X-A resin was able to significantly reduce aldehydes’ percentages: between 11% for isobutyraldehyde and 86% for phenylacetaldehyde. On the other hand, wine colour was affected and therefore the applicability of the treatment should be further studied. However, this work can be considered a starting point to solve the technological challenge involved in the elimination of aldehydes from wine.

I think I like this: assessing conscious versus subconscious wine taste responses using neuroscientific techniques

PurposeThis methodological paper aims to demonstrate the potential benefits of using consumer neuroscientific methodologies to measure consumers’ subconscious responses when consuming wine during a taste experiment. By comparing conscious and subconscious evaluations during a tasting experience this study illustrates how this methodology offers a more nuanced understanding of the consumer evaluation of wine during a consumption experience.Design/methodology/approachThe research made use of a single-case taste test experiment whereby a wine expert blind-tasted 20 white wine varietals. Throughout each tasting, subconscious responses were measured using electroencephalography (EEG), combined with conscious measures of stated preferences using a questionnaire.FindingsStark differences were observed between the results of the conscious and subconscious wine evaluation measures, underscoring the complex nature of consumer decision-making and preference development. This study practically demonstrates the use and value of EEG as a consumer neuroscientific methodology in a wine marketing context.Originality/valueThis paper demonstrates the value of neuroscience techniques in identifying differences in the conscious and subconscious wine evaluation measures. This study practically demonstrates the use and value of EEG as a consumer neuroscientific methodology in a wine marketing context.

Adapting Polarized Projective Mapping to Investigate Fruitiness Aroma Perception of White Wines from Oregon

Fruitiness is a highly desirable aroma quality in white wines. This study aimed to investigate the fruitiness aroma perception of white wines using a rapid sensory method and to compare the sensory maps obtained from wine experts and trained consumers. A polarized projective mapping (PPM) method was adapted by using fruit standards as fixed poles. Twenty-one white wines were selected for this study. Ten wine experts and twelve trained consumers participated in the adapted PPM followed by ultra-flash profiling (UFP). While experts and trained consumers used different poles to characterize the Pinot Gris and Viogniers, both panels used the stone fruit pole for the Chardonnays. UFP revealed that the Pinot Gris and Viognier samples presented fruity and floral aromas and most Chardonnay wines presented non-fruity aromas. The white wines investigated presented a wide range of fruity aromas. The use of aroma standards instead of products as poles seems to be a reliable alternative to investigate fruitiness in white wines. This study helped us to understand the predominant aromas of varietal white wines from Oregon and emphasized the importance of adapting sensory methods to investigate fruitiness perception.

Improvement of Must Fermentation from Late Harvest cv. Tempranillo Grapes Treated with Pulsed Light.

Pulsed light irradiation is a nonthermal technology currently used for the elimination of pathogens from a diverse range of food products. In the last two decades, the results obtained using PL at laboratory scale are encouraging wine experts to use it in the winemaking industry. PL can reduce native yeast counts significantly, which facilitates the use of starter cultures, reducing SO2 requirements at the same time. In this experimental set up, Tempranillo grapes were subjected to pulsed light treatment, and the fermentative performance of non-Saccharomyces yeasts belonging to the species Schizosaccharomyces pombe, Lachancea thermotolerans, Torulaspora delbrueckii, Metschnikowia pulcherrima and Hanseniaspora vineae was monitored in sequential fermentations against spontaneous fermentation and pure culture fermentation with the species Saccharomyces cerevisiae. The experimental analyses comprised the determination of anthocyanin (High performance liquid chromatography with photodiode array detector—HPLC-DAD), polyphenol index and colour (Ultraviolet-visible spectroscopy—UV-Vis spectrophotometer), fermentation-derived volatiles (Gas chromatography with flame ionization detector—GC-FID), oenological parameters (Fourier transform Infrared spectroscopy—FT-IR) and structural damage of the skin (atomic force microscopy—AFM). The results showed a decrease of 1.2 log CFU/mL yeast counts after pulsed light treatment and more rapid and controlled fermentation kinetics in musts from treated grapes than in untreated samples. The fermentations done with treated grapes allowed starter cultures to better implant in the must, although a larger anthocyanin loss (up to 93%) and an increase in hue values (1 unit) towards more yellow hues were observed for treated grapes. The development of biomass was larger in musts from treated grapes. The profile of volatile compounds and oenological parameters reveals that fermentations carried out with untreated grapes are prone to deviations from native microbiota (e.g., production of lactic acid). Finally, no severe damage on the skin was observed with the AFM on treated grapes.

Do wine experts share the same mental representation? A drawing elicitation study with wine makers, sellers, and critics

Wine experts’ mental representations have often been studied by comparing experts and novices without differentiating between the types of experts. Yet, it seemed that experts developed skills through their training and practice, and thus may have different wine representations. The first objective of this study was to examine the effect of experts’ background and actual occupation on their wine representation. Beaujolais was used as a case study as this vineyard was in constant evolution with the emergence of new mentions such as Pierres Dorées. The second objective was to evaluate how new information was integrated in experts’ mental representation. Three panels were recruited: wine makers, wine sellers, and wine critics. To access their mental representations, a drawing elicitation method was combined with an interview. The results highlighted a common structure in the representations of experts based on two poles: the first pole was linked to the environment and the vineyard, and the second pole was linked to the intrinsic aspects of wines. This universal mental representation of wine would be based on the experts' long-term memory knowledge acquired during training. Some differences linked to the experts’ profession, experience and exposure were also observed. These differences could be explained in terms of perceptual learning. Through shared experiences and exposures, experts with similar occupations and backgrounds developed similar mental representations (e.g., wine makers or wine sellers). Conversely, despite a common occupation, experts with a different experience, practice and background developed idiosyncratic wine representations (e.g., wine critics).

The sensorial experience of wine for nonexperts: How the terms frequently used in Italian guidebooks are understood by standard consumers in Vietnam

AbstractOpposites are central to many areas in the fields of Psychology and Linguistics, but they are also fundamental to the technical scales used to describe wine (e.g., the Wine and Spirit Education Trust evaluation scales). The present study explores whether it is useful to refer to opposites in order to model Vietnamese standard (vs. expert) consumers' understanding of the wine descriptors frequently used in Italian texts. Sixty‐four terms used in Product Specifications and popular Italian wine guidebooks to talk about the sensory properties of red and white wines (e.g., Hazy—Viet. Đục; Bright—Viet Sáng; Complex—Viet Nhiều hương vị; Immature—Viet Chưa ngấu, etc.) were presented to 300 Vietnamese native speakers. They were asked to select what they considered to be the opposite property. Opposites were easily found by the participants, and, interestingly, they agreed with each other on those which were the most frequently chosen. Thought‐provoking similarities and differences were revealed when these finding were compared with those of a twin study involving Italian participants.Practical ApplicationsIn order to ensure effective marketing in the wine industry and to prevent misunderstandings, it is not only important that the dimensions underlying certain terms used by experts (i.e., sommeliers, oenologists, and wine experts) are similar to those understood by nonexperts (i.e., standard consumers), but it is also vital to ascertain whether nonexperts belonging to different cultures understand the same terms in a similar way. The results emerging from the present study suggest that it may be useful to use opposites to describe the sensory properties of wine to Vietnamese standard consumers, just as it is for Italian consumers. The research resulted in a list of terms which are understood in a similar way and another list with those which are understood differently. From an applicative point of view, this may be interesting for the marketing of wine in international contexts.

How does wine ageing influence perceived complexity? Temporal-Choose-All-That-Apply (TCATA) reveals temporal drivers of complexity in experts and novices

Complexity is an important component of food and drink, especially for experience/hedonic goods such as wine. However, the contribution of the temporal aspects of the tasting experience to such top-down cognitive evaluations of wine is not well-understood. In this study, we assessed how the length of wine ageing influenced the perceived complexity and temporal evolution of flavours, as judged by wine experts (N = 22) and novices (N = 60). A total of six Madeira wines, from two grape varieties and three ageing points, were served blind. Participants first rated the complexity of each wine before using Temporal-Check-All-That-Apply (TCATA) methodology to evaluate eight taste and flavour attributes over 120 s. Finally, participants provided free-text descriptions of wine complexity. Results demonstrated that wine experts, but not novices, were able to discriminate between wines according to complexity. Moreover, TCATA results mirrored the complexity results showing experts were able to better discriminate between the samples, with older wines exhibiting different flavour trajectories and longer aftertaste durations compared to younger wines. For both novices and experts, complexity ratings were associated with the total number of attributes, flavour duration, and overall intensity, but not how much the wine changed during the tasting experience. Free-text results revealed having multiple flavours as the most important self-rated feature of wine complexity, although experts reported temporal factors more than novices. Taken together, our results indicate previously uncharted differences in how experts and novices use temporal aspects in their evaluation of wine complexity.

From Flasks to Fine Glasses: Recent Trends in Wine Economics

This short literature review on wine economics introduces the Special Issue on wine of the Italian Economic Journal. Its goal is to provide non wine experts with an overall picture of recent trends of the wine sector and of major developments of the economic literature devoted to wine markets. As the wine market deeply changed through the last decades, the first section quickly outlines these changes. The second section revise literature on wine demand while the third section is on supply and the fourth is focused on policies.

Wine experts' recognition of wine odors is not verbally mediated.

Experts have better memory for items within their domain of expertise. Critically, this does not depend on more efficient use of language. However, this conclusion is based mainly on findings from experts in visual and auditory domains. Olfactory experts constitute an interesting potential counterexample since language has been implicated to be critically involved in odor memory in previous studies. We examined the role language plays in odor recognition memory for wine experts, who typically display better wine odor memory than novices and who are also able to name odors better than lay people. This suggests wine experts' superior recognition memory for odors may be verbally mediated. In 2 experiments, recognition memory for wine odors, wine-related odors, and common odors was tested in wine experts and novices. The use of language was manipulated in Experiment 1 with an overt naming versus no-naming condition, and in Experiment 2, with a verbal interference task inhibiting covert verbalization. Across the two experiments the results showed wine experts have better recognition memory for wines, but not for wine-related or common odors, indicating their memory advantage is expertise specific. Critically, this effect was not verbally mediated, as there was no relationship between experts' ability to name wines and their memory for them. Likewise, directly inhibiting online use of verbalization did not affect memory for wine odors in experts. In sum, once expertise has been acquired, language does not play a causal role in recognition memory for odors. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Non experts’ understanding of terms frequently used by experts to describe the sensory properties of wine: An investigation based on opposites

Expert wine tasters have a greater ability than non-experts to discriminate between and evaluate the sensory properties of wine. In this paper we explore non-experts’ understanding of a set of 64 terms which are frequently used as descriptors of the sensory properties of wine. These terms can be found in guidebooks on Italian wine, in production specifications and in the official tasting scales of the Italian Sommelier Association. The 64 target terms were selected in a preliminary phase of Study 1. These terms were then presented to 558 Italian non-expert consumers who were asked to identify the opposite of each target term, in relation to red and white wines respectively. The findings a) suggest that non-experts are able to think of the sensorial properties of wine in terms of dimensions modeled by two opposite poles (and it is interesting to note that this is in line with the bipolar structure of many dimensions used by experts); b) made it possible to identify the properties that are consistently associated by non-experts with the same dimension and those which, in contrast, evoke different opposites (i.e. the underlying dimension is less univocal) and c) reveal that the target properties generally evoke similar dimensions when applied to both red and white wines. In Study 2, the gap between the experts’ interpretation of the terms and that of the non-experts was explored.Taken as a whole, the results of the study indicate an empirical methodology that may be used to investigate how standard consumers understand descriptions of wine. They may also provide food for thought regarding the ambiguity embedded in these descriptions due to potential discrepancies between the lexicons that experts and non-experts refer to in the world of wine tasting.

Is knowledge emotion? The subjective emotional responses to wines depend on level of self-reported expertise and sensitivity to key information about the wine

Many factors influence emotional responses evoked by wines. Here we assessed how self-reported wine expertise, tasting condition (blind vs. informed) as well as sensitivity to key information about wines (e.g., reputation, price, grape variety) impact the subjective affective responses they evoked. We measured subjective affective responses of high and low in self-reported wine expertise consumers to 8 different wines in a blind tasting and in a tasting when information about the wines was known. After their first tasting session, we asked participants the extent to which they considered specific information when they intended to purchase wine (e.g., reputation, etc.). The more wine consumers high in self-reported expertise paid attention to the wine’s reputation, the less they used feelings when tasting wines. In contrast, the more the wine tasters low in self-reported expertise paid attention to the wine’s reputation, the more feelings they reported. Moreover, when considering positive and negative feelings separately, it appears that the more participants paid attention to the label, the lower the number of positive terms they tended to mention. Additionally, wine tasters low in self-reported expertise were more inclined to report positive feelings towards the wines in the informed condition and if they were sensitive to wine’s reputation. In contrast, wine tasters high in self-reported expertise were less inclined to report positive feelings towards the wines in the informed condition and if they were sensitive to wine’s reputation. These results hint at the importance of considering psychological theories of emotion while studying wine-elicited emotions, in particular appraisal theories of emotion.

Access to wine experts' long-term memory to decipher an ill-defined sensory concept: the case of green red wine

The present study aims to understand an ill-defined sensory concept by a long-term memory-based strategy with Spanish winemakers from four wine regions using “green wine” as a case study. A total of 77 Spanish winemakers from four Spanish wine regions carried out a non-tasting free description task. The description task yielded terms belonging to two main categories including origin-related terms as well as sensory terms. Sensory terms belonged to aroma, taste, trigeminal, colour, multimodal and hedonic subcategories, which elucidates the multidimensionality of the studied concept. The most cited specific terms were “vegetal aroma”, “bitter” and “unpleasant”. Despite these commonalities, a certain idiosyncrasy linked to taste (“excessive sourness”) and trigeminal (“astringency”) subcategories as well as to wine components (“tannins”) was evidenced as they were cited distinctly by experts belonging to separate wine regions. The capacity of approaches based on long-term memory to decipher multidimensional and ill-defined concepts is highlighted. The regional effect is also explained in terms of cognitive processes (i.e., knowledge and experience), which is linked to the use of sensory concepts by wine experts.

Is perceptual learning generalisable in the chemical senses? A longitudinal pilot study based on a naturalistic blind wine tasting training scenario

A growing body of research has demonstrated differences in perceptual, conceptual, and language abilities between wine experts and novices. However, it is unclear to what extent these differences are innate or acquired through training. The present study assessed the olfactory and gustatory performance of a group of university blind wine tasters before and after training. Previous research has shown that this training regimen significantly improves blind tasting accuracy, but it remains unknown whether perceptual learning from blind tasting training is generalisable to standard tests of olfactory/gustatory ability. Two testing sessions were carried out for the training group (N = 14) as well as for a control group (N = 12) before and after a 5-week training period. In each session, participants underwent olfactory threshold, discrimination, and identification assessments as well as a gustatory sensitivity test. Olfactory discrimination significantly improved in the training group over the 5-week period, and the training group outperformed controls in olfactory identification in both sessions. Based on our limited set of data, wine training seems to have improved olfactory discrimination, even though the method of training did not involve odorants used in the discrimination test itself. These results reveal that even wine training over a short period seems to make concrete changes to olfactory performance, supporting the idea that generalised perceptual learning can take place for odour discrimination.

The Pricing of an Experience Good in a Competitive and Opaque Market

This paper explores the price determinants of fine wines in a competitive and opaque market. Intense international competition, high production costs and information asymmetries, result in Swiss wine prices that mostly depend on collective reputation, production techniques, and product positioning. Following an exogenous currency shock, producers have reacted with a 5% price increase. Finally, a reduction in information asymmetries with the arrival of an influential wine expert on the market had a positive price effect on rated wines and producers. These two findings suggest that wine producers have used this situation to review their wines' pricing and positioning.

“This wine is dead!”: unravelling the effect of word-of-mouth and its moderators in price-based wine quality perceptions

PurposeThe purpose is to evaluate the effect of positive and negative traditional word-of-mouth (PWOM and NWOM) on price-based quality perceptions of middle-range wine (price comprised between US$4 and US$12), considering the moderator role of type of relationship between source and receiver –in terms of the strength-of-tie – and the wine expertise of the source.Design/methodology/approachA between-subjects experimental design was employed to test the hypotheses. The dependent variable, perceived quality, was analysed using analysis of variance (ANOVA) and independent samples t-tests.Findings(1) WOM significantly affects price-based wine quality perceptions only when the source is perceived as expert, and independently on the type of relationship between source and receiver. (2) WOM has no “additive effect” on price-based quality perceptions for all but one condition (PWOM about high-priced wine from a close and expert source). (3) WOM results more useful than price to assess quality mainly in “contradictory” situations.Research limitations/implicationsThe main limitation is related to the fact that is an experiment, specifically the manipulation of strength-of-tie and source expertise. Although manipulation checks show good results for the procedure, future research should try to design better ways to manipulate these variables, or different procedures to capture similar data.Practical implicationsManagerial efforts aimed to the encouragement of PWOM will be more efficient in markets with a high proportion of experts. Also, the marketing strategy of stimulating PWOM would be more effective for wines in the low-price category.Originality/valueThis study contributes to understand the impact of WOM on wine quality perceptions by examining if WOM affects priors price-based quality perceptions. In particular, we determine if price-based quality perceptions are able to be modified by PWOM and NWOM, a “competing” approach that is novel within the wine literature in which price is usually the most used cue to elaborate quality perceptions.

Chemical, physical, and sensory attributes of Sauvignon blanc wine fermented in different kinds of vessels

The use of vessels that are alternative to conventional cylindrical stainless steel tanks and traditional oak barrels during winemaking has increased in recent years. Examples of these alternative vessels include fashionable, oval-shaped vessels or classic clay jars. A Sauvignon blanc grape juice was fermented (by using a commercial starter made of two different commercial strains from Oenobrands/Anchor Oenology) in four different vessels in triplicate: 150 L cylindrical stainless steel tanks (as a control), 980 L polyethylene oval-shaped tanks, 450 L concrete oval-shaped tanks, and 225 L clay jars. Afterwards, the finished wines were chemically, physically, and sensorially characterized. The wine fermented in the concrete vessels had a slightly higher pH (about 0.05 pH units) and lower titratable acidity (about 0.20 g/L of tartaric acid equivalents) value than those of the control wine. The wine fermented in the clay jars showed the lowest amounts of C10, C12, and C14 volatile compounds (containing about 69% of those compounds when compared with the control wine). The wine fermented in the polyethylene oval-shaped tanks was described as the most bitter, while the wine fermented in the concrete oval-shaped tanks was described as the least fruity when wines were sensory analyzed by a panel of 11 wine experts. The wines fermented in the oval-shaped vessels showed lower volatile acidity (about 25% reduction of volatile acidity when compared with wines fermented in non-oval-shaped vessels), higher residual sugars (wines fermented in the oval-shaped vessels contained about 1.7 g/L of residual sugars, while wines fermented in the non-oval-shaped vessels contained about 1.4 g/L of residual sugars), higher phosphorous content (with an increase of about 12%), and required a lower dose of bentonite to achieve protein stability (about 64 g/hL of bentonite) than those of the other wines (about 75 g/hL of bentonite). Even though some statistical differences were found among wines fermented in the different vessels, it should be noted that the differences were very small. Thus, it seems that the impact of the fermentation vessel type on final wine features is lower than expected.