Representation Of Structure Research Articles

Real-time deformable structure tracking in 3D ultrasound sequences using deformable convolutional layers.

Ultrasound imaging can provide 3D images of soft tissue structures in real-time without harmful radiation. Due to its high level of availability and low-cost characteristics, it is becoming more and more interesting for therapy guidance purposes like in radiotherapy. However, for usage in radiotherapy a robust and real-time image analysis method is required to be able to track the target during the treatment session. Soft tissue structures move in high dimensional motion patterns, including deformation especially due to breathing which makes the tracking task challenging. To overcome the deformation complexity, a novel real-time capable tracking approach in 3D ultrasound is proposed in this study. Deformable convolution layers are included in a 2D convolutional autoencoder architecture to learn deformation-invariant representations from ultrasound patches. For this, a novel 3D to 2D ultrasound patch reduction strategy is proposed. The tracking procedure is performed in the representation space of the autoencoder. We therefore implemented a greedy local search tracking algorithm and evaluated it in a preliminary study on the basis of nine expert-labeled landmarks in 18 in-vivo 3D ultrasound liver sequences. Four different autoencoder architectures with different deformable convolution arrangements are compared in the evaluation. The results show that using deformable convolution layers is beneficial compared to conventional convolution layers. A mean tracking error of 1.58 ± 0.87 mm was measured using deformable convolution, which is an improvement by 10.7% compared to conventional convolution. Additionally, evaluating the runtime shows that the tracking algorithm is real-time capable as the mean runtime per 3D ultrasound frame was around 4 ms. It could be shown that using deformable convolution layers is beneficial for learning meaningful representations of deformable structures from ultrasound patches. A tracking error similar to state-of-the-art methods could be achieved but in a runtime up to 100 times faster.

Application of mobile laser scanner to augment stationary collected point clouds

ABSTRACT Acquiring precise and comprehensive representations of architectural structures or industrial installations is crucial for a variety of monitoring and cadastral applications. A number of tools have been developed for this purpose, but they often face the problem with balancing accuracy and acquisition time. Terrestrial laser scanners typically provide the highest precision in point clouds, as they perform scanning while motionlessly mounted on rigid tripods. Careful pre-planning is needed for scanner locations on the object under investigation. Limited number of scanning locations often compromises point clouds at inaccessible locations, especially in complex environments. In this paper, we propose a solution to this issue by employing an autonomous quadrupedal robot-based laser scanner to augment the point clouds obtained with high-precision stationary scanners. We demonstrate this approach in an auditorium consisting of stairs, multi-row seats and additional complex-shaped acoustic diffusors.

An Intent Recognition Method for Aerial Swarm Based on Attention Pooling Mechanism

In the realm of unmanned aerial vehicle (UAV) swarm intent recognition, conventional approaches predominantly focus on the attributes derived from singular targets at discrete instances. This trend leads to a significant limitation: the inability to effectively harness and capture the collective feature information of the entire swarm over temporal sequences. To address this gap, this study introduces a comprehensive end-to-end UAV swarm intent recognition approach. Initially, this method utilizes the distance threat coefficient and angular threat coefficient between UAVs to construct the graphical structural representation of the UAV swarm. Subsequently, an innovative deep learning framework, designated as attention-pool based on graph attention network and long short-term memory, which integrates a graph attention network, a novel graph pooling strategy, and a long short-term memory, is developed. This architecture can process the graphically structured data derived from the swarm modeling and accurately deduce the collective intent. Through experimental validation and analyses against existing methodologies, as well as ablation studies, it is evidenced that the model outperforms state-of-the-art methods in terms of accuracy of intent recognition.

A new approach for quantification of total above-ground heartwood and sapwood volume of trees

Key messageTerrestrial laser scanning data of trees combined with models of heartwood content proportion of woody disks can provideprecise characterization of total aboveground tree sapwood and heartwood volume.Quantifying sapwood and heartwood content of trees is challenging. Previous studies have primarily characterized main stem wood composition, while branches have rarely been studied. Terrestrial Laser Scanning (TLS) can provide precise representations of the entire above-ground tree structure, non-destructively, to help estimate total tree sapwood and heartwood volume. In this study, we used TLS to scan above-ground portions of twenty-four open-grown, urban Gleditsia triacanthos trees on Michigan State University campus. TLS data were used to generate quantitative structure models that provided comprehensive characterizations of the total tree woody surface area (WSA) and volume. A subsample of trees was harvested (after scanning) and main stem and branch woody disks were collected to build models of heartwood content proportion. Models were applied to measurements from TLS to quantify complete heartwood and sapwood volume of each tree, including main stem and branches. From the base to the top of the trees, the largest portion of stem vertical cumulative volume was heartwood, whereas vertical cumulative volume of branches showed the opposite pattern. Absolute heartwood volume declined monotonically toward zero from stem base to stem top, while absolute sapwood volume declined sharply from stem base up to near the crown base and then remained relatively constant within crown. We also found that tree WSA increased with sapwood volume for both branches and main stem. This study developed a novel, general method for quantifying total aboveground sapwood and heartwood volume of trees and provided new insights into urban tree growth and structure.

How Students Use Cognitive Structures to Process Information in the Algebraic Reasoning?

Cognitive processes are procedures for using existing knowledge to combine it with new knowledge and make decisions based on that knowledge. This study aims to identify the cognitive structure of students during information processing based on the level of algebraic reasoning ability. This type of research is qualitative with exploratory methods. The data collection technique used began by providing a valid and reliable test instrument for algebraic reasoning abilities for six mathematics education student programs at the Islamic University of Sultan Agung Indonesia. Subjects were selected based on the level of upper, middle, and lower algebraic reasoning abilities. The results showed that (1) students with the highest level of algebraic reasoning ability meet the logical structure of Logical Reasoning which shows that students at the upper level can find patterns and can generalize; (2) Students at the intermediate level understand the cognitive structure of Symbolic Representations, where students can make connections between knowledge and experience and look for patterns and relationships but have difficulty making rules and generalizations; (3) students at lower levels understand the cognitive structure of Comparative Thinking, where students are only able to make connections between prior knowledge and experience.

3D Seismic Attribute Conditioning Using Multiscale Sheet-Enhancing Filtering

Seismic coherence attributes are valuable for identifying structural features, but they often face challenges due to significant background noise and non-feature-related stratigraphic discontinuities. To address this, it is necessary to apply attribute conditioning to the coherence to enhance the visibility of these structures. The primary challenge of attribute conditioning lies in finding a concise structural representation that isolates only the true interpretive features while effectively removing noise and stratigraphic interference. In this study, we choose sheet-like structures as this concise structural representation, as faults are typically characterized by their thin and narrow profiles. Inspired by multiscale Hessian-based filtering (MHF) and its application on vascular structure detection, we propose a method called anisotropic multiscale Hessian-based sheet-enhancing filtering (AMHSF). This method is specifically designed to extract and magnify sheet-like structures from noisy coherence images, with a novel enhancement function distinct from those traditionally used in vascular enhancement. The effectiveness of our AMHSF is demonstrated through experiments on both synthetic and real datasets, showcasing its potential to improve the identification of structural features in coherence images.

THE NON-COPRIME GRAPHS OF UPPER UNITRIANGULAR MATRIX GROUPS OVER THE RING OF INTEGER MODULO WITH PRIME ORDER AND THEIR TOPOLOGICAL INDICES

In its application graph theory is widely applied in various fields of science, including scheduling, transportation, industry, and structural chemistry, such as topological indexes. The study of graph theory is also widely applied as a form of representation of algebraic structures, including groups. One form of graph representation that has been studied is non-coprime graphs. The upper unitriangular matrix group is a form of group that can be represented in graph form. This group consists of upper unitriangular matrices, which are a special form of upper triangular matrix with entries in a ring R and all main diagonal entries have a value of one. In this research, we look for the form of a non-coprime graph from the upper unitriangular matrix group over a ring of prime modulo integers and several topological indexes, namely the Harmonic index, Wiener index, Harary index, and First Zagreb index. The findings of this research indicate that the structure of the graph and the general formula for the Harmonic index, Wiener index, Harary index, and First Zagreb index were successfully obtained.

Large-scale-aware data augmentation for reduced-order models of high-dimensional flows

Convolutional autoencoders have proven to be an adequate tool to perform reduced-order modeling for high-dimensional nonlinear dynamical systems. Their goal is to reduce dimensionality strongly while preserving the most characteristic features of the system. Here, we show that these models rely sensitively on the completeness of the provided data. This is particularly challenging for fully turbulent flows with their coherent structures ranging from large-scale superstructures to dissipative eddies over orders of magnitude in time and space. As a result, an unrealistically large number of data snapshots would be required to properly cover all the essential dynamics, whereas the features on small time and length scales require only a small number of snapshots of the respective flow, especially the long lasting large-scale structures that are difficult to characterize either numerically or experimentally. We demonstrate for three types of flows that a missing representation of large-scale turbulent structures leads to failures in the training process. We suggest a method to mitigate this shortcoming. This includes the transformation of data samples to new large-scale structures, which enhance the data. Furthermore, we skip augmentations that are more detrimental to the model performance. We evaluate our method for three datasets, two from numerical simulations of turbulent Rayleigh–Bénard convection flows and one from laboratory experiment for the flow past an array of cylinders. We show that the method can substantially improve model utility for high-dimensional data. In this way, we avoid an intensive grid search through possible augmentation combinations without further knowledge about the underlying system.

Systematic Review and Framework for AI-Driven Tacit Knowledge Conversion Methods and Machine Learning Algorithms for Ontology-Based Chatbots in E-Learning Platforms

The conversion of tacit knowledge, which is deeply rooted in personal experience and often difficult to articulate, presents a significant challenge within knowledge management systems. Ontology-based chatbots offer a promising solution by leveraging structured knowledge representations and advanced natural language processing (NLP) techniques to facilitate this transformation. This paper explores the various methods and algorithms used in developing ontology-based chatbots, with a particular focus on their role in converting tacit knowledge into more accessible forms. Additionally, it provides a comparative analysis of the algorithms employed, highlighting their respective strengths and weaknesses. Ultimately, this study addresses the critical challenge of managing and converting tacit knowledge, with the aim of enhancing the overall effectiveness of knowledge management systems.

Cortical encoding of spatial structure and semantic content in 3D natural scenes.

Our visual system enables us to effortlessly navigate and recognize real-world visual environments. Functional magnetic resonance imaging (fMRI) studies suggest a network of scene-responsive cortical visual areas, but much less is known about the temporal order in which different scene properties are analysed by the human visual system. In this study, we selected a set of 36 full-colour natural scenes that varied in spatial structure and semantic content that our male and female human participants viewed both in 2D and 3D while we recorded magnetoencephalography (MEG) data. MEG enables tracking of cortical activity in humans at millisecond timescale. We compared the representational geometry in the MEG responses with predictions based on the scene stimuli using the representational similarity analysis framework. The representational structure first reflected the spatial structure in the scenes in time-window 90-125 ms, followed by the semantic content in time-window 140-175 ms after stimulus onset. The 3D stereoscopic viewing of the scenes affected the responses relatively late, from around 140 ms from stimulus onset. Taken together, our results indicate that the human visual system rapidly encodes a scene's spatial structure and suggest that this information is based on monocular instead of binocular depth cues.Significance statement Our visual system enables us to recognize and navigate our visual surroundings seemingly effortlessly, but what exactly happens in our brains remains poorly understood. With the help of time-resolved brain imaging (magnetoencephalography, MEG), we found that the brain first encodes the spatial structure of a scene (e.g., cluttered or navigable) before its semantic content (e.g., a car park or farm). Brain imaging studies typically use 2D pictures as stimuli. Here we asked whether binocular disparity, a depth cue which arises from our two eyes seeing the scene from slightly different angles, aids the coding of the spatial structure. Our results suggest that this 3D depth cue plays little role in the rapid, initial sensing of our spatial surroundings.

A Symbolic AI Approach to Medical Training

In traditional medical education, learners are mostly trained to diagnose and treat patients through supervised practice. Artificial Intelligence and simulation techniques can complement such an educational practice. In this paper, we present GLARE-Edu, an innovative system in which AI knowledge-based methodologies and simulation are exploited to train learners “how to act” on patients based on the evidence-based best practices provided by clinical practice guidelines. GLARE-Edu is being developed by a multi-disciplinary team involving physicians and AI experts, within the AI-LEAP (LEArning Personalization of AI and with AI) Italian project. GLARE-Edu is domain-independent: it supports the acquisition of clinical guidelines and case studies in a computer format. Based on acquired guidelines (and case studies), it provides a series of educational facilities: (i) navigation, to navigate the structured representation of the guidelines provided by GLARE-Edu, (ii) automated simulation, to show learners how a guideline would suggest to act, step-by-step, on a specific case, and (iii) (self)verification, asking learners how they would treat a case, and comparing step-by-step the learner’s proposal with the suggestions of the proper guideline. In this paper, we describe GLARE-Edu architecture and general features, and we demonstrate our approach through a concrete application to the melanoma guideline and we propose a preliminary evaluation.

Grammar-constrained decoding for structured information extraction with fine-tuned generative models applied to clinical trial abstracts.

In the field of structured information extraction, there are typically semantic and syntactic constraints on the output of information extraction (IE) systems. These constraints, however, can typically not be guaranteed using standard (fine-tuned) encoder-decoder architectures. This has led to the development of constrained decoding approaches which allow, e.g., to specify constraints in form of context-free grammars. An open question is in how far an IE system can be effectively guided by a domain-specific grammar to ensure that the output structures follow the requirements of a certain domain data model. In this work we experimentally investigate the influence of grammar-constrained decoding as well as pointer generators on the performance of a domain-specific information extraction system. For this, we consider fine-tuned encoder-decoder models, Longformer and Flan-T5 in particular, and experimentally investigate whether the addition of grammar-constrained decoding and pointer generators improve information extraction results. Toward this goal, we consider the task of inducing structured representations from abstracts describing clinical trials, relying on the C-TrO ontology to semantically describe the clinical trials and their results. We frame the task as a slot filling problem where certain slots of templates need to be filled with token sequences occurring in the input text. We use a dataset comprising 211 annotated clinical trial abstracts about type 2 diabetes and glaucoma for training and evaluation. Our focus is on settings in which the available training data is in the order of a few hundred training examples, which we consider as a low-resource setting. In all our experiments we could demonstrate the positive impact of grammar-constrained decoding, with an increase in F 1 score of pp 0.351 (absolute score 0.413) and pp 0.425 (absolute score 0.47) for the best-performing models on type 2 diabetes and glaucoma datasets, respectively. The addition of the pointer generators had a detrimental impact on the results, decreasing F 1 scores by pp 0.15 (absolute score 0.263) and pp 0.198 (absolute score 0.272) for the best-performing pointer generator models on type 2 diabetes and glaucoma datasets, respectively. The experimental results indicate that encoder-decoder models used for structure prediction for information extraction tasks in low-resource settings clearly benefit from grammar-constrained decoding guiding the output generation. In contrast, the evaluated pointer generator models decreased the performance drastically in some cases. Moreover, the performance of the pointer models appears to depend both on the used base model as well as the function used for aggregating the attention values. How the size of large language models affects the performance benefit of grammar-constrained decoding remains to be more structurally investigated in future work.

Screening Cyclodextrin Complexes for Bisphenols with High Binding Performance Based on the Data-Driven Model.

The distinctive cavity structure of cyclodextrin, which results in binding properties, is credited with its application prospects in chemical, pharmacy, and material fields. The binding capacity can be regulated by substituting the hydroxyl groups on the cyclodextrins. It is possible to acquire anticipated binding properties by designing the modified groups on cyclodextrins. In this article, a data-driven model is proposed with a novel cyclodextrin/guest structure representation method to assist the cyclodextrin design. The model's performance is verified via several validations, as the squared correlation coefficients for cross-validation (Q2) and test set (R2test) are 0.801 and 0.841, respectively. With the proposed model and fluorescence experiments for cyclodextrin/bisphenol complexes, several cyclodextrin hosts, which have a strong binding capacity for bisphenols, are screened, synthesized, and characterized. The results show a controlled average absolute error of 0.605 M-1, suggesting the feasibility of data supplementation and molecular design. It is believed that the data-driven model can serve as theoretical assistance and a driving tool for the cyclodextrin complexes design, potentially leading to advancements in cyclodextrin's industrial applications and scientific research.

Mapping Mental Representations With Free Associations: A Tutorial Using the R Package associatoR.

People's understanding of topics and concepts such as risk, sustainability, and intelligence can be important for psychological researchers and policymakers alike. One underexplored way of accessing this information is to use free associations to map people's mental representations. In this tutorial, we describe how free association responses can be collected, processed, mapped, and compared across groups using the R package associatoR. We discuss study design choices and different approaches to uncovering the structure of mental representations using natural language processing, including the use of embeddings from large language models. We posit that free association analysis presents a powerful approach to revealing how people and machines represent key social and technological issues.

Mathematical modeling and QSPR analysis of hepatitis treatment drugs through connection indices an innovative approach.

Chemical structures may be defined based on their topology, which allows for the organization of molecules and the representation of new structures with specific properties. We use topological indices, which are precise numerical measurements independent of structure, to measure the bonding arrangement of a chemical network. An essential objective of studying topological indices is to collect and alter chemical structure data to develop a mathematical relationship between structures and physico-chemical properties, bio-activities, and associated experimental factors. Topological indices establish the correlation between molecular characteristics, such as physical, chemical, thermodynamic, and biological activity, and their corresponding chemical structures in Quantitative Structure-Property Relationships (QSPR) and Quantitative Structure-Activity Relationships (QSAR). Hepatitis, in its advanced stages, can lead to the development of mental illnesses. Effective management of symptoms frequently entails regular use of medicine and therapy for a prolonged period. Assurance of the safety and efficacy of drug design is of utmost importance. Various parameters, including solubility, metabolic stability, toxicity, permeability, and transporter effects, depending on the physical and chemical properties of the treatment, impact the efficacy of biopharmacological design. Increasingly, computational approaches have become indispensable in the discipline of pharmaceutical discoveries and development. The major objective of this work is to examine the chemical appropriateness of connection indices in evaluating six physico-chemical properties of the 14 medicines used for the treatment of hepatitis. This work conducts QSPR analysis to obtain the most accurate approximations for the physico-chemical attributes of pharmacological agents employed in the treatment of hepatitis.

Label-free imaging flow cytometry for cell classification based directly on multiple off-axis holographic projections.

Imaging flow cytometry allows highly informative multi-point cell analysis for biological assays and medical diagnosis. Rapid processing of the imaged cells during flow allows real-time classification and sorting of the cells. Off-axis holography enables imaging flow cytometry without chemical cell staining but requires digital processing to the optical path delay profile for each frame before the cells can be classified, which slows down the overall processing throughput. We present a method for real-time cell classification via label-free quantitative imaging flow cytometry using digital holography, offering a comprehensive representation of cellular structures, without the need for digital processing before automatic cell classification. We aim to develop an automatic cell classification scheme based directly on the off-axis holographic projections of the cells during flow and test it for stain-free imaging flow cytometry of white blood cells. After building a dedicated off-axis holographic microscopy system for acquiring white blood cells during flow, we apply deep-learning classification directly in the off-axis hologram space, rather than in the quantitative phase profile space. This way, we simplify computational processes and allow a significant increase in the cell classification throughput. In addition, by utilizing multiple-viewpoint holographic projections of the cells rotated during flow, instead of using a single projection, we obtain better classification results due to the additional cellular information gained. Our technique demonstrates increasing accuracy with additional viewpoint holographic projections from the optical system, achieving a 7.69% improvement when processing ten interferometric projections compared with a single interferometric projection (regular off-axis hologram). Our technique also outperforms using multiple optical path delay profile projections, requiring off-axis holographic digital preprocessing, by 17.95%, because the holographic projections are analyzed directly without preprocessing and includes the amplitude information as well. Our cell classification approach has great potential for high-throughput, high-content, label-free imaging flow cytometry for classification of large-scale cellular datasets and real-time cell classification during flow in clinical settings.

Design, development, and evaluation of the organic chemistry representational competence assessment (ORCA)

This paper describes the design and evaluation of the O̲rganic chemistry R̲epresentational C̲ompetence A̲ssessment (ORCA). Grounded in Kozma and Russell's representational competence framework, the ORCA measures the learner's ability to interpret, translate, and use six commonly used representations of molecular structure (condensed structures, Lewis structures, skeletal structures, wedge-dash diagrams, Newman projections, and chair conformations). Semi-structured interviews with 38 first-semester organic chemistry learners informed the development of the ORCA items. The ORCA was developed and refined through three pilot administrations involving a total of 3477 first-semester organic chemistry students from multiple institutions. The final version of the ORCA was completed by 1494 students across five institutions. Various analyses provided evidence for the validity and reliability of the data generated by the assessment. Both one-factor and three-factor correlated structures were explored via confirmatory factor analysis. The one-factor model better captured the underlying structure of the data, which suggests that representational competence is better evaluated as a unified construct rather than as distinct, separate skills. The ORCA data reveal that the representational competence skills are interconnected and should consistently be reinforced throughout the organic chemistry course.

A Topology-Enhanced Multi-Viewed Contrastive Approach for Molecular Graph Representation Learning and Classification.

In recent times, graph representation learning has been becoming a hot research topic which has attracted a lot of attention from researchers. Graph embeddings have diverse applications across fields such as information and social network analysis, bioinformatics and cheminformatics, natural language processing (NLP), and recommendation systems. Among the advanced deep learning (DL) based architectures used in graph representation learning, graph neural networks (GNNs) have emerged as the dominant and highly effective framework. The recent GNN-based methods have demonstrated state-of-the-art performance on complex supervised and unsupervised tasks at both the node and graph levels. In recent years, to enhance multi-view and structured graph representations, contrastive learning-based techniques have been developed, introducing models known as graph contrastive learning (GCL) models. These GCL approaches leverage unsupervised contrastive methods to capture multi-view graph representations by comparing node and graph embeddings, yielding significant improvements in both graph-level representations and task-specific applications, such as molecular embedding and classification. However, as most GCL techniques are primarily designed to focus on the explicit graph structure through GNN-based encoders, they often overlook critical topological insights that could be provided through topological data analysis (TDA). Given the promising research indicating that topological features can greatly benefit various graph learning tasks, we propose a novel topology-enhanced, multi-view graph contrastive learning model called TMGCL. Our TMGCL model is designed to capture and utilize both comprehensive multi-scale topological and global structural information from graphs. This enhanced representation capability positions TMGCL to directly support a range of applications, such as molecular classification, with improved accuracy and robustness. Extensive experiments within two real-world datasets proved the effectiveness and outperformance of our proposed TMGCL in comparing with state-of-the-art GNN/GCL-based baselines.

Building Detection and Outlining in Multi-Modal Remote Sensor Data: A Stratified Approach

Exploitation of aerial sensor data for vectorized representation of urban structures is of great importance for urban modeling. Even if elevation data is available, extracting complex and irregular building shapes would become challenging. In this paper, we present a stratified approach for the extraction of building outlines consisting of two steps: Classification and vectorization. For classification, we use training data transfer to evaluate a dataset with no or little labeled data. Since the available reference data is also flawed, we use a self-developed interactive tool to adjust and improve the building shape before contrasting it with the classification results. Initial building polygons are slightly generalized and refined considering building shape characteristics. Hereby, Least-Squares adjustment is implemented to solve the best-fit problem of building edges with the input data by applying the Gauss-Helmert and Gauss-Markov Models. On the raster level, the resulting polygons achieved an accuracy of over 99% in the Potsdam dataset and almost 98% in the Munich dataset while on the vector level, the median building-wise deviations lie in sub-meter range. Although there were few mis-detections and, sometimes, considerable peak deviations for the Hausdorff distance, the compelling qualitative results attest to the robustness and validity of the proposed procedure.

STOUT V2.0: SMILES to IUPAC name conversion using transformer models

Naming chemical compounds systematically is a complex task governed by a set of rules established by the International Union of Pure and Applied Chemistry (IUPAC). These rules are universal and widely accepted by chemists worldwide, but their complexity makes it challenging for individuals to consistently apply them accurately. A translation method can be employed to address this challenge. Accurate translation of chemical compounds from SMILES notation into their corresponding IUPAC names is crucial, as it can significantly streamline the laborious process of naming chemical structures. Here, we present STOUT (SMILES-TO-IUPAC-name translator) V2, which addresses this challenge by introducing a transformer-based model that translates string representations of chemical structures into IUPAC names. Trained on a dataset of nearly 1 billion SMILES strings and their corresponding IUPAC names, STOUT V2 demonstrates exceptional accuracy in generating IUPAC names, even for complex chemical structures. The model's ability to capture intricate patterns and relationships within chemical structures enables it to generate precise and standardised IUPAC names. While established deterministic algorithms remain the gold standard for systematic chemical naming, our work, enabled by access to OpenEye’s Lexichem software through an academic license, demonstrates the potential of neural approaches to complement existing tools in chemical nomenclature.Scientific contribution STOUT V2, built upon transformer-based models, is a significant advancement from our previous work. The web application enhances its accessibility and utility. By making the model and source code fully open and well-documented, we aim to promote unrestricted use and encourage further development.Graphical