Harnessing Food Product Reviews for Personalizing Sweetness Levels

Abstract

Sweet taste is innately appealing, ensuring that mammals are attracted to the sweetness of mother’s milk and other sources of carbohydrates and calories. In the modern world, the availability of sugars and sweeteners and the eagerness of the food industry to maximize palatability, result in an abundance of sweet food products, which poses a major health challenge. The aim of the current study is to analyze sweetness levels, liking, and ingredients of online reviews of food products, in order to obtain insights into sensory nutrition and to identify new opportunities for reconciling the palatability–healthiness tension. We collected over 200,000 reviews of ~30,000 products on Amazon dated from 2002 to 2012 and ~350,000 reviews of ~2400 products on iHerb from 2006 to 2021. The reviews were classified and analyzed using manual curation, natural language processing, and machine learning. In total, ~32,000 (Amazon) and ~29,000 (iHerb) of these reviews mention sweetness, with 2200 and 4600 reviews referring to the purchased products as oversweet. Oversweet reviews were dispersed among consumers. Products that included sucralose had more oversweet reviews than average. 26 products had at least 50 reviews for which at least 10% were oversweet. For these products, the average liking by consumers reporting oversweetness was significantly lower (by 0.9 stars on average on a 1 to 5 stars scale) than by the rest of the consumers. In summary, oversweetness appears in 7–16% of the sweetness-related reviews and is less liked, which suggests an opportunity for customized products with reduced sweetness. These products will be simultaneously healthier and tastier for a substantial subgroup of customers and will benefit the manufacturer by expanding the products’ target audience. Analysis of consumers’ reviews of marketed food products offers new ways to obtain informative sensory data.

Natural Language Processing for Analyzing Online Customer Reviews: A Survey, Taxonomy, and Open Research Challenges

“…The quick customer loyalty model for e-commerce [88] and the Level of Success model (LOS) [134] also recognize the need to turn sentiment analytics into customer satisfaction and retention efforts. The use of NLP to extract insights from user-generated reviews in nutraceutical retail [87] and the study of sweetness in online food product reviews [146] demonstrate the multifaceted uses of NLP in customer satisfaction. Implementing NLP with recommendation systems shows its importance in user experiences and purchase decisions.…”

“…Analyzing Amazon and iHerb reviews, the research on sweetness in food products identified opportunities for less sweet products catering to a healthier consumer base. The study employed manual curation, NLP, and machine learning to reveal the impact of sweetness on product liking, suggesting potential benefits for health-conscious customers and manufacturers [146]. Challenging the belief that longer product reviews are uniformly more helpful, the study utilized advanced machine learning methods to analyze Amazon reviewsʹ sentence-level argumentation.…”

Natural Language Processing for Analyzing Online Customer Reviews: A Survey, Taxonomy, and Open Research Challenges

“…Analyzing Amazon and iHerb reviews, the research on sweetness in food products identified opportunities for less sweet products catering to a healthier consumer base. The study employed manual curation, NLP, and machine learning to reveal the impact of sweetness on product liking, suggesting potential benefits for health-conscious customers and manufacturers [146]. Challenging the belief that longer product reviews are uniformly more helpful, the study utilized advanced machine learning methods to analyze Amazon reviews' sentence-level argumentation.…”