Domain Specificity Research Articles

A GraphKAN-Based Intelligent Fault Diagnosis Method of Rolling Bearing Under Variable Working Conditions

Unsupervised domain adaptation (UDA) can effectively address the two main drawbacks of transfer learning: the requirement of a large number of samples collected from different working conditions, and the inherent defects of convolutional neural networks (CNNs). In the realm of UDA, it is essential to leverage three types of information: class labels, domain specifications, and data organization. These components play a vital role in linking the source domain with the target domain. A technique aimed at identifying issues in rolling bearings is presented, employing an integration of CNN-KAN and GraphKAN structures to support the UDA methodology. A cohesive deep learning architecture is employed to represent the three types of information involved in UDA. The initial two types of information are represented through the roles of classifier and domain discriminator. To begin with, an architecture leveraging CNN-KAN is employed to extract features from the incoming signals. Following this, the features obtained from the CNN-KAN architecture are input into a specially developed graph creation layer that constructs instance graphs by analyzing the relationships among the structural characteristics found within the samples. In the following step, an innovative GraphKAN model is applied to illustrate the instance graphs, concurrently employing CORrelation ALignment (CORAL) loss to assess the structural discrepancies among instance graphs from different domains. Results from experiments conducted on two separate datasets demonstrate that the proposed framework surpasses alternative approaches and successfully recognizes transferable characteristics that are advantageous for domain adaptation.

Exploring the Domain Specificity and the Neural Correlates of Memory Unawareness in Alzheimer's disease

Exploring the Domain Specificity and the Neural Correlates of Memory Unawareness in Alzheimer's disease

Characterization of the host specificity of the SH3 cell wall binding domain of the staphylococcal phage 88 endolysin.

Bacteriophages produce endolysins at the end of the lytic cycle, which are crucial for lysing the host cells and releasing virion progeny. This lytic feature allows endolysins to act as effective antimicrobial alternatives when applied exogenously. Staphylococcal endolysins typically possess a modular structure with one or two enzymatically active N-terminal domains (EADs) and a C-terminal cell wall binding domain (CBD). The EADs degrade the peptidoglycan layer, leading to bacterial lysis, while the CBD binds to the specific host cell wall, and therefore, influences specificity of the endolysin. This study aimed to alter and characterize the host specificity of the CBD by exploring the impact of amino acid modifications within the CBD of a staphylococcal endolysin, Endo88. Endo88 was able to lyse Staphylococcus spp. and Enterococcus faecalis. However, despite attempts to mutate amino acids hypothesized for binding with cell wall components, the host-range was not affected but the lytic activity was severely reduced instead, although no alterations were performed on the EADs (Cysteine, histidine-dependent aminohydrolases/peptidases domain and Amidase domain). Further investigations of the CBD alone (Src homology3 domain, SH3) without the EADs suggested that binding and lytic activity may not be correlated in some cases since Endo88 and its mutants could lyse Staphylococcus epidermidis well but no binding activity was observed in the flow cytometry analysis. Molecular docking was used to gain insights on the observations for the binding and lytic activity which may help future strategies in designing enhanced engineered endolysins.

510. Bacteriophages as Drug Delivery Vehicles for Polymyxin B

Abstract Background The goal of this project is to develop a safer, more effective formulation of polymyxin B for E. coli bacteremia. Polymyxins are a family of cyclic lipopeptides that display exceptional antibiotic activities toward gram-negative pathogens such as E. coli, Pseudomonas aeruginosa, and Acinetobacter baumannii. The PMB label includes an FDA black box warning for serious nephrotoxicity and neurotoxicity. However, today, with the growing emergence of “superbugs” and a dry antibiotic discovery pipeline, PMB is increasingly used as a last-line therapy to treat systemic infections. Thus, although PMB is a potent antibacterial agent, its toxicity currently prevents its use as an earlier or more widespread treatment. Transmission electron microscopy (TEM; negative stain) images of anti-LPS phage bound to gram-negative organisms. Phages were labeled with gold nanoparticles (dark spheres) coated with antibodies against pVIII, allowing phages to be easily identified (red darts). Gold-labeled anti-LPS phage are shown (A) without bacteria; or incubated with (B) E. coli DH5α; (C) E. coli ATCC BAA 1161; (D) A. baumannii ATCC 19606. Phage width is increased by the labeling reagents. Scale bars = 500 nm. Methods We engineered M13 to bind the core antigen of lipopolysaccharide (LPS), enabling binding to a variety of top-priority pathogenic gram-negative bacterial species, including E. coli, P. aeruginosa, A. baumannii, Klebsiella pneumoniae, and Burkholderia cepacia. Since the anti-LPS phages are able to bind but do not infect (nonviable), they are not cytotoxic by themselves. However, delivery of PMB by the anti-LPS phage confers antibiotic activity. Thus, in our strategy, PMB molecules are conjugated to the anti-LPS phage capsid, and delivered specifically to gram-negative bacterial cells through the specificity and high affinity of the anti-LPS binding domain. We have shown that the PMB – anti-LPS phage conjugate (termed “PMB–Phage” hereafter) effectively lowers the minimal inhibitory concentration of PMB by multiple orders of magnitude in vitro compared to PMB sulfate. MIC and MBC of PMB and PMB-Phage, determined in vitro for several gram-negative organisms. The heat map corresponds to the MIC or MBC values. Equivalent PMB concentrations for PMB-Phage MIC and MBC were calculated from the measured phage concentration and the average amount of PMB loaded per virion. * = resistant. ND = not determined due to technical limitation on phage concentration. Results We have already demonstrated following results.A broad-spectrum anti-LPS phage was engineered and verified to bind to several gram-negative bacterial species including E. coli.Polymyxin B was conjugated to the anti-LPS phage, creating PMB-Phage, which was highly antibacterial in vitro.PMB-Phage is non-toxic to mammalian cells in vitro and non-toxic in mice.PMB-Phage is distributed to various organs within 1 hour and has an elimination half-life of approximately 1 day.PMB-Phage appears to be effective in treating E. coli bacteremia in a mouse model. PMB-Phage Biodistribution Figure 9: Time course biodistribution of PMB-Phage in various organs, over time in male mice with (red) or without (blue) 1×10^7cfu (non-lethal dose) of E. coli ATCC BAA 1161 injected (n=3). Signals are measured as a percentage of the injected dose (ID) per volume (per cc). Representative biodistribution images (j) of 1 animal. Conclusion PMB-Phage will greatly reduce the nephrotoxicity of PMB, enabling safe treatment for a broader patient population. Liver and kidney blood biomarkers at end point of 7-day toxicity studies. Animals in experimental groups were injected 100µL of test materials through IV tail vein injection daily for 7 consecutive days. Experimental group animals for each condition (N=3), receiving either PMB or PMB-Phage at different doses show no significant difference comparing to control groups of animals group receiving no treatment (N=5) and 1xPBS buffer (N=3), in light and dark grey columns respectively. The only exception being all animals receiving 72 mg/kg body weight PMB (N=3) did not survive a single dose and died on day 1 (indicated by red cross in figure). Disclosures All Authors: No reported disclosures

Scaling BERT for Healthcare: An End-to-End Framework for Medical Document Automation

This article proposes a comprehensive framework for implementing BERT-based language models at scale for automated medical document processing in healthcare environments. Recent studies have demonstrated BERT models achieving accuracy rates of up to 89.7% in medical entity recognition tasks [2], suggesting significant potential for healthcare applications. The proposed architecture introduces a novel hierarchical attention mechanism specifically engineered to capture the nested complexity of medical documentation while maintaining computational efficiency in production systems. Drawing inspiration from successful implementations in pharmacy settings [15, 16], this framework features a distributed training pipeline designed to process annotated medical documents across multiple specialties, coupled with a dynamic medical vocabulary injection system that aims to preserve BERT's contextual understanding while potentially reducing false positives in entity recognition. By leveraging distributed computing infrastructure and optimized model architecture, the framework could theoretically achieve substantial improvements over traditional approaches in both accuracy and processing efficiency, particularly in handling complex medical terminology and context-sensitive information extraction. The proposed system architecture anticipates addressing current challenges in healthcare documentation, where systems process an average of 6.2 million clinical documents annually [1]. This theoretical framework contributes to the field by proposing a scalable methodology for implementing advanced language models in healthcare settings while addressing the unique challenges of medical domain specificity and production deployment requirements.

Dense Video Captioning: A Survey of Techniques, Datasets and Evaluation Protocols

Untrimmed videos have interrelated events, dependencies, context, overlapping events, object-object interactions, domain specificity, and other semantics that are worth highlighting while describing a video in natural language. Owing to such a vast diversity, a single sentence can only correctly describe a portion of the video. Dense Video Captioning (DVC) aims to detect and describe different events in a given video. The term DVC originated in the 2017 ActivityNet challenge, after which considerable effort has been made to address the challenge. Dense Video Captioning is divided into three sub-tasks: (1) Video Feature Extraction (VFE), (2) Temporal Event Localization (TEL), and (3) Dense Caption Generation (DCG). In this survey, we discuss all the studies that claim to perform DVC along with its sub-tasks and summarize their results. We also discuss all the datasets that have been used for DVC. Lastly, current challenges in the field are highlighted along with observatory remarks and future trends in the field.

Method for Creating Domain-Specific Dataset Ontologies from Text in Uncontrolled English

Abstract Automated understanding of activities in enterprises is challenging due to a lack of domain specifications and a lack of domain ontologies. The goal of this research is to develop a method to extract elements of domain-specific processes from textual documents in unstructured English and form domain dataset ontologies. In order to achieve the goal, the related work on discourse analysis and business process modelling have been considered. The prominent technologies for implementation of the proposed method are machine learning, including classification algorithms and natural language processing using a large language model. The first experimental results are presented, and further research is discussed. Potentially, the method proposed can be implemented as a part of some assisting tool for system analysts and can support an analysis of the domain-specific information by providing contextual information from this and potentially related domains.

Exploiting the inherent promiscuity of the acyl transferase of the stambomycin polyketide synthase for the mutasynthesis of analogues.

The polyketide specialized metabolites of bacteria are attractive targets for generating analogues, with the goal of improving their pharmaceutical properties. Here, we aimed to produce C-26 derivatives of the giant anti-cancer stambomycin macrolides using a mutasynthesis approach, as this position has been shown previously to directly impact bioactivity. For this, we leveraged the intrinsically broad specificity of the acyl transferase domain (AT12) of the modular polyketide synthase (PKS), which is responsible for the alkyl branching functionality at this position. Feeding of a panel of synthetic and commercially available dicarboxylic acid 'mutasynthons' to an engineered strain of Streptomyces ambofaciens (Sa) deficient in synthesis of the native α-carboxyacyl-CoA extender units, resulted in six new series of stambomycin derivatives as judged by LC-HRMS and NMR. Notably, the highest product yields were observed for substrates which were only poorly accepted when AT12 was transplanted into a different PKS module, suggesting a critical role for domain context in the overall functioning of PKS proteins. We also demonstrate the superiority of this mutasynthesis approach - both in terms of absolute titers and yields relative to the parental compounds - in comparison to the alternative precursor-directed strategy in which monoacid building blocks are supplied to the wild type strain. We further identify a malonyl-CoA synthetase, MatB_Sa, with specificity distinct from previously described promiscuous enzymes, making it a useful addition to a mutasynthesis toolbox for generating atypical, CoA activated extender units. Finally, we show that two of the obtained (deoxy)-butyl-stambomycins exhibit antibacterial and antiproliferative activities similar to the parental stambomycins, while an unexpected butyl-demethyl congener is less potent. Overall, this works confirms the interest of biosynthetic pathways which combine a dedicated route to extender unit synthesis and a broad specificity AT domain for producing bioactive derivatives of fully-elaborated complex polyketides.

Fractional order control of Wind Energy Conversion System based on DFIG

The paper deals with the elaboration of a robust and efficient decoupled active and reactive powers control algorithm of double-fed induction generator (DFIG) using fractional order control based on proportional and integral controllers. First, the models of wind turbine and electrical generator are established, as well as a decoupled active and reactive powers control of DFIG. Second, the used fractional order controllers are designed using an analytical design method based on frequency domain specifications, namely phase margin, robustness to gain variations and gain limitation at the crossover frequency. Third, to evaluate the fractional order controller control performances, simulation results are presented and discussed for all the operating region of the WECS. Then, simulation results are presented and interpreted considering a variable wind speed in order to reach all the operating regions of the WECS. Finally, a conclusion is given in the light of the obtained results.

THE CRITICAL YET OVERLOOKED SPATIAL COMPETENCE IN LEARNING ASTRONOMY: DECODING SEMANTIC SPATIAL INFORMATION IN PICTURES

This study examined the impact of spatial ability on science achievement by distinguishing between domain-general and domain-specific spatial abilities, focusing on their relationship with the conceptual understanding of the apparent path of the sun. Additionally, the mediating effects of domain-specific spatial knowledge on the relationships between spatial ability and science achievement were tested. A mixed-methods design was used. Forty university students participated. Mental rotation and astronomical drawing tests were used to evaluate their general spatial ability and science achievement. The domain-specific spatial knowledge used in solving astronomical drawing tests was ascertained via interviews and analyzed, with results presented as static-type semantic spatial propositions and dynamic-type spatial skills. The results demonstrated an overlap between general spatial ability and domain-specific spatial knowledge, challenging the traditional dichotomy between them. Furthermore, domain-specific spatial knowledge fully mediated the relationship between general spatial ability and science achievement. These findings have significant implications for astronomy education, highlighting the importance of domain-specific spatial competence over general ability in solving domain problems and underscoring the need for explicit instruction in the decoding of semantic spatial information in pictorial representations, which is often overlooked in studies focusing on spatial skills training or the surface characteristics of scientific representations. Keywords: spatial ability, domain generality, domain specificity, semantic spatial knowledge, astronomy education

The Neural Specificity of Interference Resolution in Phonological, Semantic, and Visual Domains at Different Ages.

The question of whether cognitive control is specific to certain domains or domain-general remains an extensively debated question at both cognitive and neural levels. This study examined the neural substrates associated with resistance to interference (RI) in phonological, semantic, and visual domains by using strictly matched tasks and determining the domain-general or domain-specific manner in which aging affects the neural substrates associated with RI. In an fMRI experiment, young and older participants performed a similarity judgment task with phonological, semantic, or visual interference buildup. For both age groups, domain-specific RI effects were observed at the univariate level, with increased involvement in the phonological domain of the right angular gyrus and the right lingual gyrus, in the semantic domain of the bilateral inferior frontal gyrus, the bilateral superior parietal and angular gyri and the left middle temporal gyrus, and in the visual domain of the middle/superior frontal gyri and occipital gyri. At the multivariate level, although RI effects could be decoded from neural patterns in the bilateral inferior frontal gyrus for all domains and age groups, between-domain prediction of RI conditions was associated with Bayesian evidence for the null hypothesis. This study supports the domain specificity of neural substrates associated with RI while stressing its age independency.

The human Dicer helicase domain is capable of ATP hydrolysis and single-stranded nucleic acid binding

BackgroundVertebrates have one Dicer ortholog that generates both microRNAs (miRNAs) and small interfering RNAs (siRNAs), in contrast to the multiple Dicer-like proteins found in flies and plants. Here, we focus on the functions of the human Dicer (hDicer) helicase domain. The helicase domain of hDicer is known to recognize pre-miRNA substrates through interactions with their apical loop regions. Besides interacting with canonical substrates, the hDicer helicase domain has also been suggested to bind many different cellular RNAs; however, a comprehensive study of the biochemical activities and substrate specificity of the hDicer helicase domain towards different nucleic acids has yet to be undertaken.ResultsHere, we reveal that full-length hDicer, through its helicase domain, hydrolyzes ATP. The ATPase activity of hDicer can only be observed under low-turnover conditions. To the best of our knowledge, this is the first time this activity has been reported for vertebrate Dicers. We also show that the hDicer helicase domain binds single- but not double-stranded RNAs and DNAs and that this binding activity presumably is not nucleotide-dependent. Moreover, the hDicer helicase domain may influence the structure of the RNA to which it binds.ConclusionsPreservation of ATPase activity by hDicer suggests that this enzyme performs many more functions in the cell than is currently assumed. Our findings open new avenues for future studies aimed at defining the cellular activities of hDicer that may be associated with these newly described biochemical properties: ATP hydrolysis and single-stranded nucleic acid binding activities.

Development of a PET Probe Targeting Bromodomain and Extra-Terminal Proteins for In Vitro and In Vivo Visualization.

Background: Bromodomain and extra-terminal (BET) proteins are critical regulators of gene transcription, as they recognize acetylated lysine residues. The BD1 bromodomain of BRD4, a member of the BET family, has emerged as a promising therapeutic target for various diseases. This study aimed to develop and evaluate a novel C-11 labeled PET radiotracer, [11C]YL10, for imaging the BD1 bromodomain of BRD4 in vivo. Methods: [11C]YL10 was synthesized and evaluated for its ability to bind to the BD1 bromodomain selectively. PET imaging studies were conducted in mice to assess brain penetration, pharmacokinetics, and selectivity. In vitro autoradiography and blocking experiments were performed to confirm the tracer's specificity for the BD1 domain. Results: [11C]YL10 demonstrated good brain penetration, high selectivity for the BD1 bromodomain, and favorable pharmacokinetics in initial PET imaging studies. In vitro autoradiography and blocking experiments confirmed the specific binding of [11C]YL10 to the BD1 domain of BRD4, further validating its potential as a targeted radiotracer. Conclusions: The development of [11C]YL10 provides a new tool for studying BRD4 bromodomains using PET imaging technology. This radiotracer offers potential advancement in the diagnosis and research of neurodegenerative diseases and related disorders involving BRD4 dysregulation.

Deep mutationally scanned (DMS) CHIKV E3/E2 virus library maps viral amino acid preferences and predicts viral escape mutants of neutralizing CHIKV antibodies.

As outbreaks of chikungunya virus (CHIKV), a mosquito-borne alphavirus, continue to present public health challenges, additional research is needed to generate protective and safe vaccines and effective therapeutics. Prior research has established a role for antibodies in mediating protection against CHIKV infection, and the early appearance of CHIKV-specific IgG or IgG neutralizing antibodies protects against progression to chronic CHIKV disease in humans. However, the importance of epitope specificity for these protective antibodies and how skewed responses contribute to development of acute and chronic CHIKV-associated joint disease remains poorly understood. Here, we describe the deep mutational scanning of one of the dominant targets of neutralizing antibodies during CHIKV infection, the E3/E2 (also known as p62) glycoprotein complex, to simultaneously test thousands of p62 mutants against selective pressures of interest in a high throughput manner. Characterization of the virus library revealed achievement of high diversity while also selecting out non-functional virus variants. Furthermore, this study provides evidence that this virus library system can comprehensively map sites critical for the neutralization function of antibodies of both known and unknown p62 domain specificities.

A generalized network with domain invariance and specificity representation for bearing remaining useful life prediction under unknown conditions

A generalized network with domain invariance and specificity representation for bearing remaining useful life prediction under unknown conditions

Rational design of potent phosphopeptide binders to endocrine Snk PBD domain by integrating machine learning optimization, molecular dynamics simulation, binding energetics rescoring, and in vitro affinity assay.

Human Snk is an evolutionarily conserved serine/threonine kinase essential for the maintenance of endocrine stability. The protein consists of a N-terminal catalytic domain and a C-terminal polo-box domain (PBD) that determines subcellular localization and substrate specificity. Here, an integrated strategy is described to explore the vast structural diversity space of Snk PBD-binding phosphopeptides at a molecular level using machine learning modeling, annealing optimization, dynamics simulation, and energetics rescoring, focusing on the recognition specificity and motif preference of the Snk PBD domain. We further performed a systematic rational design of potent phosphopeptide ligands for the domain based on the harvested knowledge, from which a few potent binders were also confirmed by fluorescence-based assays. A phosphopeptide PP17 was designed as a good binder with affinity improvement by 6.7-fold relative to the control PP0, while the other three designed phosphopeptides PP7, PP13, and PP15 exhibit a comparable potency with PP0. In addition, a basic recognition motif that divides potent Snk PBD-binding sequences into four residue blocks was defined, namely [Χ-5Χ-4]block1-[Ω-3Ω-2Ω-1]block2-[pS0/pT0]block3-[Ψ+1]block4, where the X represents any amino acid, Ω indicates polar amino acid, Ψ denotes hydrophobic amino acid, and pS0/pT0 is the anchor phosphoserine/phosphothreonine at reference residue position 0.

Partial prior transfer learning based on self-attention CNN for EEG decoding in stroke patients

The utilization of motor imagery-based brain-computer interfaces (MI-BCI) has been shown to assist stroke patients activate motor regions in the brain. In particular, the brain regions activated by unilateral upper limb multi-task are more extensive, which is more beneficial for rehabilitation, but it also increases the difficulty of decoding. In this paper, self-attention convolutional neural network based partial prior transfer learning (SACNN-PPTL) is proposed to improve the classification performance of patients’ MI multi-task. The backbone network of the algorithm is SACNN, which accords with the inherent features of electroencephalogram (EEG) and contains the temporal feature module, the spatial feature module and the feature generalization module. In addition, PPTL is introduced to transfer part of the target domain while preserving the generalization of the base model while improving the specificity of the target domain. In the experiment, five backbone networks and three training modes are selected as comparison algorithms. The experimental results show that SACNN-PPTL had a classification accuracy of 55.4%±0.17 in four types of MI tasks for 22 patients, which is significantly higher than comparison algorithms (P < 0.05). SACNN-PPTL effectively improves the decoding performance of MI tasks and promotes the development of BCI-based rehabilitation for unilateral upper limb.

Gritty Parenting: The Development and Validation of the Parental Grit Scale.

In this study, we aimed to expand the domain specificity of grit by exploring it in the realm of parenting. Parental grit was defined as parents' inclination to demonstrate perseverance, sustained passion, and adaptability in pursuing long-term parenting goals. With longitudinal data from 1,373 U.S. parents, we developed and validated the Parental Grit Scale (PGS) in three phases: item development and revision, establishment of factor structure, and examination of criterion-related and incremental validity. The PGS exhibited a three-factor structure and (partial) scalar measurement invariance across genders, the status of the child's special needs, and measurement occasions. The PGS demonstrated moderate temporal stability. We found evidence of its validity in predicting parental emotional exhaustion, self-efficacy, autonomy support, and parents' perceptions of children's school well-being. Parental grit explained variance in these outcomes beyond domain-general grit and conscientiousness. The PGS has sound psychometric properties and is a valuable measure in the parenting domain.

Transactional development of science and mathematics knowledge and reading proficiency for multilingual students across languages of instruction.

This longitudinal study explored the reciprocal relations between students' domain-specific knowledge (science and mathematics) and reading proficiency from kindergarten to Grade 5. We compared these relational trajectories across both domains in the overall sample and examined the domain specificity of these relations within a multilingual subsample with varying language instruction backgrounds. Using latent curve modeling with structured residuals on a nationally representative data set, we identified two key patterns. In the overall sample, higher reading proficiency at kindergarten was associated with greater growth in science and mathematics knowledge, with a particularly pronounced effect in science. The predictive power of science knowledge on reading proficiency strengthened significantly from Grades 2 to 5, while reciprocal relations in mathematics intensified over time. For multilingual students, outcomes varied by the language of instruction. Those receiving English-only instruction showed early correlations between science and mathematics knowledge and reading proficiency; however, initial science and mathematics knowledge did not predict long-term growth in reading proficiency. Conversely, multilingual students who received instruction in their native language showed no immediate correlations at kindergarten. Nonetheless, their early science and mathematics knowledge significantly predicted later growth in reading proficiency. The findings underscore the critical role of native-language instruction in providing an accessible, vital cognitive and linguistic foundation that supports deeper domain knowledge building, highlighting the enduring benefits of native-language scaffolding. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Chemosensory vocabulary in wine, perfume and food product reviews: Insights from language modeling

Chemosensory sensations are often hard to describe and quantify. Language models may facilitate a systematic understanding of sensory descriptions. We accessed consumer and expert reviews of wine, perfume, and food products (English language; about 68 million words in total) and analyzed their sensory descriptions. Using a novel data-driven method based on natural language data, we compared the three chemosensory vocabularies (wine, perfume, food) with respect to their vocabulary overlap and semantic properties, and explored their semantic spaces. The three vocabularies primarily differ with respect to domain specificity, concreteness, descriptor type preference and degree of gustatory vs. olfactory association. Wine vocabulary primarily distinguishes between white wine and red wine flavors and qualities. Food vocabulary separates drinkable and edible food products and ingredients, on the one hand, and savory and non-savory products, on the other. A salient distinction in all three vocabularies is between concrete and abstract/evaluative terms. Valence also plays a role in the semantic spaces of all three vocabularies, but valence is less prominent here than ingeneral olfactory vocabulary. Our method allows a systematic comparison of sensory descriptors in the three product domains and provides a data-driven approach to derive sensory lexicons that can be applied by sensory scientists.