Defying Flavor Limits: Tackling Taste in Low Salt, Low Sugar, and Alcohol-Free Products

Leverage Knowledge Graphs and trigeminal interactions for healthier products
Written by Paddy Sudhakar - 14.06.2023

How interdisciplinary approaches integrating diverse aspects from the Knowledge Graph can aid in the discovery of gap-filling sensory properties towards the development of healthier products without compromising flavor and taste preferences. 

1.Current NPD landscape for products with improved health footprints 

1.1 Roadblocks and Challenges

Have you ever had the chance to taste products with low sugar, salt, fat, or products derived entirely from plant-based sources? And ever noticed why such products, which are usually considered as healthy, are in general poor in real or perceived flavor?  

Recent scientific and marketing studies on these emerging food trends reported a trend towards reduced liking by consumers compared to conventional products. Therefore, real or perceived loss of flavor-related sensations or sensory gaps is a significant bottleneck for the success of products with improved health footprints.  

This is the current challenge of  companies focussing on the development of novel and innovative food and beverage products. 


1.2 Solutions

One of the ways by which we at Foodpairing address this challenge is by identifying “natural” ingredients which have sensory properties comparable with popular ingredients and/or have novel sensory properties. 

The perception of flavor is modulated by inputs not only from olfaction but also from gustation and trigeminal somatosensation.  Among the major chemosensory systems in humans, the trigeminal somatosensory system plays a key role in mediating the flavor perception of foods and beverages.

Hence, by identifying ingredients or combinations of ingredients which provide additional sensory inputs such as trigeminal sensations, sensory gaps can be compensated and thereby improve the overall flavor and taste perception leading to better and higher acceptance of innovative products by consumers.  

2. Tools and Resources for customer-centric NPD of innovative products


To tackle the challenges of developing innovative products within our NPD pipelines, we use the following resources/methodologies which feed into each other. 

2.1 Novel ingredients, compounds

As discussed before, the inclusion of ingredients with novel trigeminal sensations eases the development of innovative products with improved health and/or environmental impacts. Towards this, at Foodpairing, we strive to profile and document as many novel ingredients as possible from diverse cultural and culinary traditions. We have a dedicated team with the expertise to quantitatively assign compositional data (molecules, aromas) and sensory attributes to ingredients and products. 

2.2 Knowledge Graph:

The success of innovative products depends on their conformity to various criteria ranging from matching flavor profiles, trigeminal sensations, popularity, trendiness, novelty, impacts on health and environment, likeability by the consumer target group of the ingredients to state a few. In order to evaluate the above criteria, we need heterogeneous data sources and a framework to store and retrieve them for thousands of ingredients.

At Foodpairing, we have deployed a Knowledge Graph (KG) that integrates and semantically interrelates the different types and categories of data, but also captures various multi-level hierarchies of related entities, often referred to as taxonomies. Such hierarchical structures organize knowledge in a top-down fashion, with broader categories at the top and more specific classes beneath them, e.g., product categories and sub-categories, types of sensory descriptors, etc.

2.3 Robust Real-world approaches for optimizing consumer satisfaction:

Perhaps the most important part of the NPD cycle is the optimization of consumer satisfaction. We have a diverse range of tools and methodologies which identify why a particular consumer likes or dislikes a particular product for example. This enables us to precisely target various consumer groups with tailored ingredient combinations and products, thereby maximizing consumer satisfaction. These approaches, which are often supplemented by surveys, also provide an indication as to what the consumers are looking for in innovative products, both from a sensory/taste/flavor and value perspective.  

3. Case study: Reduced alcoholic drinks

To demonstrate the practical value of our integrated approach to NPD, we briefly outline a case study using the knowledge-graph pertaining to sensates for the formulation of non-alcoholic drinks. 

Sensates are molecules which are naturally occurring in various ingredients and/or food products and are able to induce specific sensations. Sensations are triggered by the interactions between the sensates with particular receptor proteins in the trigeminal cavities (Figure 1).  

defying flavor low salt

Figure 1. Graphical representation of the knowledge-graph derived relationships between sensation-inducing ligands and trigeminal receptors. Node size is proportional to its connectivity. Edge size is proportional to the number of peer-reviewed scientific papers supporting the interaction between the ligand and the receptor.  Sensates (ethanol, gingerol, shogaol, zingerone) and their receptors (TRPA1, TRPM8, TRPV1) relevant to the case study are indicated by their common names/symbols. 


Using the schema of our multi-faceted and multi-layered KG (Figure 2),  various attributes and characteristics of the molecules and/or the ingredients which contain these molecules can be extracted. 

Figure 2. Schema of the FoodPairing Knowledge Graph. 

For example, the molecules/ligands can be classified according to the sensations they elicit in a receptor dependent or independent manner. 

Figure 3. Distribution of sensations, as inferred from the FoodPairing Knowledge Graph, elicited by molecular ligands. 

Ethanol, the primary constituent in commercially available alcoholic drinks, primarily induces the sensations of burning, bitterness and tingling mainly by activating the receptors TRPV1 and TRPA1 (Figure 4). 

By mining our Knowledge graph for ingredients containing molecules which could elicit almost the same sensations, we inferred that ginger-derived molecules such as gingerol, zingerone and shogaol can mimic the sensory effects of ethanol (Figure 5). Thus compounds from actual ingredients familiar in cuisines can be used as alternatives to either substitute ethanol in non-alcoholic drinks or reduce the concentration of ethanol in non-alcoholic drinks.   

Figure 4. Sensation induced by ethanol and the molecular mechanisms mediating them. Green edges – stimulation; Red edges – inhibition; Edge size is proportional to the number of peer-reviewed scientific papers supporting the interaction between the ligand and the receptor.

Figure 5. Sensation induced by the ginger-derived compounds (shogaol, gingerol and zingerone) and the molecular mechanisms mediating them. Green edges – stimulation; Edge size is proportional to the number of peer-reviewed scientific papers supporting the interaction between the ligand and the receptor.

4. Conclusion and Outlook

Sensory gaps in flavor perception are one of the main bottlenecks for the market breakthrough of innovative products with low salt, sugar and alcohol content. Such gaps can be filled by combining sensory data with biological mechanistic data among others. By leveraging the diverse types of datasets (Figure 2) within our knowledge graph, we at FoodPairing are in a unique position to collaborate with and help R&D teams of FMCG companies dealing with existing NPD challenges in the market such as formulating innovative foods and drinks with improved health footprints. 


Want to know how to successfully reformulate products?

Want to know how Foodpairing can help you launch successful products by keeping your consumer at the center of the development process?