Entire Dataset Research Articles

Filtering coupled Wright-Fisher diffusions.

Coupled Wright-Fisher diffusions have been recently introduced to model the temporal evolution of finitely-many allele frequencies at several loci. These are vectors of multidimensional diffusions whose dynamics are weakly coupled among loci through interaction coefficients, which make the reproductive rates for each allele depend on its frequencies at several loci. Here we consider the problem of filtering a coupled Wright-Fisher diffusion with parent-independent mutation, when this is seen as an unobserved signal in a hidden Markov model. We assume individuals are sampled multinomially at discrete times from the underlying population, whose type configuration at the loci is described by the diffusion states, and adapt recently introduced duality methods to derive the filtering and smoothing distributions. These respectively provide the conditional distribution of the diffusion states given past data, and that conditional on the entire dataset, and are key to be able to perform parameter inference on models of this type. We show that for this model these distributions are countable mixtures of tilted products of Dirichlet kernels, and describe their mixing weights and how these can be updated sequentially. The evaluation of the weights involves the transition probabilities of the dual process, which are not available in closed form. We lay out pseudo codes for the implementation of the algorithms, discuss how to handle the unavailable quantities, and briefly illustrate the procedure with synthetic data.

Automated classification of online reviews of otolaryngologists.

The study aimed to extract online comments of otolaryngologists in the 20 most populated cities in the United States from healthgrades.com, develop and validate a natural language processing (NLP) logistic regression algorithm for automated text classification of reviews into 10 categories, and compare 1- and 5-star reviews in directly-physician-related and non-physician-related categories. 1977 1-star and 12,682 5-star reviews were collected. The primary investigator manually categorized a training dataset of 324 1-star and 909 5-star reviews, while a validation subset of 100 5-star and 50 1-star reviews underwent dual manual categorization. Using scikit-learn, an NLP algorithm was trained and validated on the subsets, with F1 scores evaluating text classification accuracy against manual categorization. The algorithm was then applied to the entire dataset with comparison of review categorization according to 1- and 5-star reviews. F1 scores for NLP validation ranged between 0.71 and 0.97. Significant associations emerged between 1-star reviews and treatment plan, accessibility, wait time, office scheduling, billing, and facilities. 5-star reviews were associated with surgery/procedure, bedside manner, and staff/mid-levels. The study successfully validated an NLP text classification system for categorizing online physician reviews. Positive reviews were found to have an association with directly-physician related context. 1-star reviews were related to treatment plan, accessibility, wait time, office scheduling, billing, and facilities. This method of text classification effectively discerned the nuances of human-written text, providing valuable insights into online healthcare feedback that is scalable.Level of evidence: Level 3.

Translanguaging in instruction and pedagogic design: promoting learner-centredness in a teacher-centred reading classroom of a Bangladeshi university

ABSTRACT This study investigated the impact of translanguaging pedagogy on the reading comprehension process of first-year students studying at an English department of a Bangladeshi public university. Through a two-pronged ethnographic research design, data were collected through classroom observation, pedagogical intervention, a semi-structured interview with the class teacher and a focus group discussion with students. The entire data set was analyzed using ‘Moment Analysis,’ a conceptual framework, to provide pedagogical evidence supporting how translanguaging in instruction can facilitate students embracing a more active and collaborative role in a teacher-centered classroom. This study suggests that translanguaging-oriented instruction and pedagogic design can stimulate metalinguistic awareness to improve comprehension while enhancing active engagement with the reading materials to the benefit of the students who do not have English as a first language. Furthermore, translanguaging pedagogies can foster individualism (agency), a quality that English departments promote among their students. The findings have implications for planning future policy, including interim stages, that can bridge the English language proficiency gap between pre-tertiary and tertiary education in Bangladesh.

Comparative Analysis of Water Applicability Predictions Explained by The LightGBM Model Using SHAP and LIME

Abstract. This article mainly discusses the application of the LightGBM model to predict water potability for a dataset containing multiple water quality features. The focus of the study is to use two popular model interpretability techniques: SHAP and LIME to explain the model's prediction results. The results show that SHAP can globally explain the feature importance of the entire dataset and provide a deep understanding of the features and model behavior, while LIME provides a detailed explanation of a single prediction through local linear approximation, which is easier to interpret and apply. This article also compares the strengths and limitations of SHAP and LIME in explaining the LightGBM models behavior, demonstrating their applicability and explanatory power in different contexts. In addition, this article also explores the actual application scenarios of water quality prediction and analyzes how interpretability improves model transparency and trust in this field. Through these analyses, the article provides practical suggestions on how to choose appropriate model interpretation methods in reality.

Metaphorical meaning dynamics: Identifying patterns in the metaphorical evolution of English words using mathematical modeling techniques

Abstract Conceptual metaphor theory has been criticized due to its emphasis on concepts instead of words and its top-down direction of analysis. In response to these criticisms, this paper employs a new strategy, utilizing established mathematical modeling methods to allow a systematic, quantitative analysis of the entire dataset produced by the Mapping Metaphor project at the University of Glasgow. This dataset consists of 9609 words performing 18916 metaphorical mappings between 414 domains. The data is represented as a network consisting of 414 nodes, the domains, connected by shared words. Words are represented by groups of directed mappings between all domains in which they occur. This is made possible by the use of a directed hypergraph representation, a tool commonly used in discrete mathematics and various areas of computer science but not previously applied to the metaphorical meanings of words. Examining the dataset as a whole, rather than focusing on individual words or metaphors, allows global patterns of behavior to emerge from the data without pre-filtering or selection by the authors. Outcomes of the analysis relating to the distributions of source and target domains within the network, the growth mechanisms at work in the spread of metaphorical meanings and how these relate to existing concepts in CMT are discussed.

Applying a Support Vector Machine (SVM-RFE) Learning Approach to Investigate Students’ Scientific Literacy Development: Evidence from Asia, Europe, and South America

Cultivating scientific literacy is a goal widely shared by educators and students around the world. Many studies have sought to enhance students’ proficiency in scientific literacy through various approaches. However, there is a need to explore the attributes associated with advanced levels of scientific literacy, especially the influence of contextual factors. In this context, our study employs a machine learning technique—the SVM-RFE algorithm—to identify the critical characteristics of students with strong scientific literacy in Asia, Europe, and South America. Our research has pinpointed 30 key factors from a broader set of 162 contextual factors that are indicative of outstanding scientific literacy among 15-year-old secondary school students. By utilizing student samples from the three continents, our study provides a comprehensive analysis of these factors across the entire dataset, along with a comparative examination of the optimal set of key factors between continents. The findings highlight the importance of these key factors, which should be considered by educational policymakers and school leaders when developing educational policies and instructional strategies to foster the most effective development of scientific literacy.

Prediction of Dry Mouth Condition Using Radiomics Features from Tongue Diagnosis Image

Xerostomia, commonly known as dry mouth, is characterized by reduced salivary secretion, which can lead to various oral health issues and discomfort. In this paper, we propose a novel, non-invasive method for predicting xerostomia through the analysis of tongue images. To predict salivary gland secretion from tongue images, we collected images from patients who visited the hospital with complaints of dry mouth and measured their saliva secretion. Features were extracted from these tongue images, and correlation analysis was performed using machine learning techniques to assess the relationship between the extracted features and measured saliva secretion. We obtained tongue images and saliva secretion measurements from 176 patients. Images were cropped to 100 × 100 pixels, resulting in 462 features. The dataset was divided into training and test sets, consisting of 160 and 16 samples, respectively. The correlation coefficients for the training and test datasets were 0.9496 and 0.9415, respectively, while the correlation coefficient for the entire dataset was 0.9482. The estimated linear equation was y = 0.9244x + 2.1664. This study aimed to predict salivary gland secretion based on tongue images. By extracting features from color images and employing a neural network machine learning model, we estimated salivary gland secretion. With a sufficiently large dataset of tongue images, further advancements in regression analysis using deep learning techniques could enhance the accuracy of these predictions.

A Descriptive Comparison of the Epidemiological Characteristics of Delta and Omicron Variant-Driven Outbreaks in Bhutan

SARS-CoV-2 rapidly mutated, causing different waves of outbreaks worldwide. Bhutan experienced three major outbreaks of COVID-19 before experiencing the outbreak driven by the Omicron variant in January 2022. The data collected by the National Outbreak Investigation and Surveillance Team during the Delta variant-driven outbreak and Omicron outbreak were accessed and analyzed. The data were analyzed using R statistical software. Descriptive analysis was carried out for the entire dataset and the statistical comparison between the two outbreaks was carried out using student’s t-test and Pearson’s chi-square test. During the Delta variant-driven outbreak, a total of 1648 cases were reported, with a daily average of 13 cases. The highest one-day case number reported was 99. On the contrary, within 33 days, a total of 3788 cases were reported with a daily average of 115 cases during the Omicron outbreak. The highest one-day case number reported was 312. The median incubation period was 3 days (range = 0–18 days) and 1 day (range = 0–14 days) during the Delta and Omicron-driven outbreaks, respectively. The number of symptomatic cases was significantly higher during the Delta outbreak (p-value < 0.0001). Of the total cases reported during the Delta outbreak, 1175 (71.3%) had received a single dose of the vaccine, 79 (5%) received two doses, while 394 (24%) were unvaccinated. During the Omicron outbreak, 1957 (52%) cases had received their booster (third dose), 904 (23.8%) received two doses and only 40 (1%) received a single dose of the vaccine. The number of unvaccinated cases was 887 (23.4%), of which 375 (10%) were children below 12 years. Our findings corroborate the enhanced transmissibility of the Omicron variant as reported elsewhere. We report significantly less symptomatic cases during the Omicron outbreak. Further, our data show that the incubation period for the Omicron variant is shorter compared to the Delta variant (p-value < 0.0001).

Shared oscillatory mechanisms of alpha-band activity in prefrontal regions in eyes open and closed state using a portable EEG acquisition device

Alpha oscillations are associated with various psychiatric disorders, with many studies focusing on the prefrontal cortex, where transcranial alternating current stimulation (TACS) is applied in the alpha frequency band. This approach often involves selecting individualized alpha frequencies to resonate with their endogenous alpha oscillations. While strong alpha oscillations (8–13 Hz) are typically induced when the eyes are closed, they can also occur during the resting state with eyes open. However, it remains unclear whether these alpha oscillations share a common neural generation mechanism. Exploring which of these alpha oscillations is more suitable as a stable alpha peak frequency is a question of significant interest. Therefore, to systematically study this issue, we specifically collected resting-state electroencephalographic (EEG) data from the prefrontal region of 40 individuals, under both eyes-open and closed- eye conditions, with multiple follow-ups extending up to nine months. Through spectral analysis on each person’s entire dataset and averaging the results, we observed a significant positive correlation between the alpha-band power in the eyes-open and the eyes-closed states, in terms of both absolute power and relative power. Further analysis revealed that this correlation was primarily contributed by the periodic activity within the alpha band. Upon modelling the oscillatory components, we discovered distinct differences in the oscillatory characteristics-such as number of the alpha sub-oscillations between the eyes-open state and the eyes-closed state. Our study is the first to systematically explored the relationship between alpha oscillations in the prefrontal cortex in the eyes-open and eyes-closed states, identifying both shared part of the neural generation mechanism and some distinct neural mechanisms that are unique to each state.

An Intraoperative Ultrasound Evaluation of Axillary Lymph Nodes: Cassandra Predictive Models in Patients with Breast Cancer—A Multicentric Study

Background and Objectives: Axillary lymph node (ALN) staging is crucial for the management of invasive breast cancer (BC). Although various radiological investigations are available, ultrasound (US) is the preferred tool for evaluating ALNs. Despite its immediacy, widespread use, and good predictive value, US is limited by intra- and inter-operator variability. This study aims to evaluate US and Elastosonography Shear Wave (SW-ES) parameters for ALN staging to develop a predictive model, named the Cassandra score (CS), to improve the interpretation of findings and standardize staging. Materials and Methods: Sixty-three women diagnosed with BC and treated at two Italian hospitals were enrolled in the study. A total of 529 lymph nodes were surgically removed, underwent intraoperative US examination, and were individually sent for a final histological analysis. The study aimed to establish a direct correlation between eight US-SWES features (margins, vascularity, roundness index (RI), loss of hilum fat, cortical thickness, shear-wave elastography hardness (SWEH), peripheral infiltration (PI), and hypoechoic appearance) and the histological outcome (benign vs. malignant). Results: Several statistical models were compared. PI was strongly correlated with malignant ALNs. An ROC analysis for Model A revealed an impressive AUC of 0.978 (S.E. = 0.007, p < 0.001), while in Model B, the cut-offs of SWEH and RI were modified to minimize the risk of false negatives (AUC of 0.973, S.E. = 0.009, p < 0.001). Model C used the same cut-offs as Model B, but excluded SWEH from the formula, to make the Cassandra model usable even if the US machine does not have SW-ES capability (AUC of 0.940, S.E. = 0.015, p < 0.001). A two-tiered model was finally set up, leveraging the strong predictive capabilities of SWEH and RI. In the first tier, only SWES and RI were evaluated: a positive result was predicted if both hardness and roundness were present (SWES > 137 kPa and RI < 1.55), and conversely, a negative result was predicted if both were absent (SWES < 137 kPa and RI > 1.55). In the second tier, if there was a mix of the results (SWES > 137 kPa and RI > 1.55 or SWES < 137 kPa and RI < 1.55), the algorithm in Model B was applied. The model demonstrated an overall prediction accuracy of 90.2% in the training set, 87.5% in the validation set, and 88.9% across the entire dataset. The NPV was notably high at 99.2% in the validation set. This model was named the Cassandra score (CS) and is proposed for the clinical management of BC patients. Conclusion: CS is a simple, non-invasive, fast, and reliable method that showed a PPV of 99.1% in the malignancy prediction of ALNs, potentially being also well suited for young sonographers.

Integrating Multisource Data and Machine Learning for Supraglacial Lake Detection: Implications for Environmental Management and Sustainable Development Goals in High Mountainous Regions

The accurate detection and monitoring of supraglacial lakes in high mountainous regions are crucial for understanding their dynamic nature and implications for environmental management and sustainable development goals. In this study, we propose a novel approach that integrates multisource data and machine learning techniques for supra-glacial lake detection in the Passu Batura glacier of the Hunza Basin, Pakistan. We extract pertinent features or parameters by leveraging multisource datasets such as radar backscatter intensity VH and VV parameters from Sentinel-1 Ground Range Detected (GRD) data, near-infrared (NIR), NDWI_green, NDWI_blue parameters from Sentinel-2 Multi-spectral Instrument(MSI) data, and surface slope, aspect, and elevation parameters from topographic data. The entire dataset is partitioned into training and testing sets, with machine learning models including the artificial neural network (ANN), the support vector machine (SVM), logistic regression (LR), random forest (RF), and K-nearest neighbour (KNN) trained on the training data (70%). Accuracy assessment employs testing data and involves the evaluation of metrics such as ROC curves and confusion matrices. The best-performing model, ANN, is validated against manually digitized lake polygons derived from Sentinel-2 and Google Earth Pro imagery. Furthermore, the digitized lake polygons are used to analyze glacial lake dynamics from 2016 to 2022. Key findings of this research presented that the NDWI_green, Sigma0_VH, and elevation are the most significant predictors in detecting supra-glacial lakes. Among the various trained and evaluated models, the Artificial Neural Network (ANN) achieved the highest performance (accuracy: 95%, AUC: 0.99) and accurately mapped supra-glacial lakes regardless of their small size. The findings have significant implications for understanding glacial lake behavior in the context of climate change and informing future research and monitoring efforts.

Quantification of Fundus Autofluorescence Features in a Molecularly Characterized Cohort of More Than 3500 Inherited Retinal Disease Patients from the United Kingdom

PurposeTo quantify relevant fundus autofluorescence (FAF) image features cross-sectionally and longitudinally in a large cohort of patients with inherited retinal diseases (IRDs). DesignRetrospective study of imaging data (55-degree blue-light FAF, Heidelberg Spectralis) . ParticipantsPatients with a clinical and molecularly confirmed diagnosis of IRD who have undergone 55-degree FAF imaging at Moorfields Eye Hospital (MEH) and the Royal Liverpool Hospital (RLH) between 2004 and 2019. MethodsFive FAF features of interest were defined: vessels, optic disc, perimacular ring of increased signal (ring), relative hypo-autofluorescence (hypo-AF) and hyper-autofluorescence (hyper-AF). Features were manually annotated by six graders in a subset of patients based on a defined grading protocol to produce segmentation masks to train an AI model, AIRDetect, which was then applied to the entire imaging dataset. Main Outcome MeasuresQuantitative FAF imaging features including area in mm2 and vessel metrics, were analysed cross-sectionally by gene and age, and longitudinally to determine rate of progression. AIRDetect feature segmentation and detection were validated with Dice score and precision/recall, respectively. ResultsA total of 45,749 FAF images from 3,606 IRD patients from MEH covering 170 genes were automatically segmented using AIRDetect. Model-grader Dice scores for disc, hypo-AF, hyper-AF, ring and vessels were respectively 0.86, 0.72, 0.69, 0.68 and 0.65. Across patients within our cohort at presentation, the five genes with the largest hypo-AF areas were CHM, ABCC6, RDH12, ABCA4, and RPE65, with mean per-patient areas of 43.72, 29.57, 20.07, 19.65, and 16.92 mm2. The five genes with the largest hyper-AF areas were BEST1, CDH23, NR2E3, MYO7A, and RDH12, with mean areas of 0.50, 047, 0.44, 0.38, and 0.33 mm2 respectively. The five genes with largest ring areas were NR2E3, CDH23, CRX, EYS and PDE6B, with mean areas of 3.60, 2.90, 2.89, 2.56, and 2.20 mm2. Vessel density was found to be highest in EFEMP1, BEST1, TIMP3, RS1, and PRPH2 (11.0%, 10.4%, 10.1%, 10.1%, 9.2%) and was lower in Retinitis Pigmentosa (RP) and Leber Congenital Amaurosis genes. Longitudinal analysis of decreasing ring area in four RP genes (RPGR, USH2A, RHO, EYS) found EYS to be the fastest progressor at -0.18 mm2/year. ConclusionsWe have conducted the first large-scale cross-sectional and longitudinal quantitative analysis of FAF features across a diverse range of IRDs using a novel AI approach.

Impact of time window on estimating representativeness error in sea surface salinity data

ABSTRACT Validating satellite data using buoys presents a significant challenge due to the differing spatial and temporal representativeness of these data sources. This discrepancy introduces two critical issues in satellite data validation: the selection of the matchup window during the collocation process and the estimation of representativeness error. While numerous studies have addressed these issues independently, research on their interaction remains limited. Utilizing quadruple Sea Surface Salinity (SSS) matchup datasets, this study investigates the impact of the time window on the estimation of SSS representativeness error between two satellite observations: SMOS and SMAP. Our results reveal a clear dependence of representativeness error on the selected time windows. In the global ocean, the variance of representativeness errors within a small time window is 0.040 psu2, which is substantially greater than that of the entire dataset (0.016 psu2). In contrast, the representativeness errors within larger time windows are systematically lower than those within smaller time windows. This research also finds that this dependency is not influenced by the randomness of the matchup data, climatological Sea Surface Salinity, or geographic locations. Instead, our findings suggest that the impact of the time window on representativeness error may reflect the characteristics of satellite data across different time intervals. Satellites are more likely to capture SSS signals within smaller time windows, resulting in larger representativeness errors when comparing satellite SSS with buoy measurements. Conversely, larger time windows lead to significant mismatches between satellite observations and hinder the ability of satellites to detect SSS signals, which can result in smaller or even reversed representativeness errors.

A systematic study of sheep faeces to be used as an alternative inoculum source in the Hohenheim gas test

ABSTRACT In vitro studies of gas production (GP) using rumen fluid obtained from rumen-cannulated animals are common in feed evaluation for ruminants. The objective was to compare sheep faeces with rumen fluid from dairy cows as inocula in the Hohenheim gas test (HGT) using a large number of different-quality feeds. It then was evaluated whether GP obtained by using rumen inoculum (RI-GP) can be estimated from GP using faecal inoculum (FI-GP). Ninety feeds for ruminants including roughages (n = 27), total mixed rations (n = 6), commercial compound feeds for dairy cows (n = 18), cereal grains (n = 10), energy-rich by-products (n = 5), legume grains (n = 6), oilseed meals (n = 6), and other protein-rich feeds (n = 12) were incubated with either rumen inoculum (RI) or faecal inoculum (FI), and GP was recorded at 2, 4, 8, 12, 24, 48, and 72 h. Nonlinear equations were fitted to the GP data to determine GP kinetic parameters. Correlations between RI-GP and FI-GP were calculated for different time points. Linear regressions were calculated to estimate RI-GP from FI-GP using the entire data set or subsets of roughages and concentrates. GP kinetic parameters were lower for all feed categories when incubated with FI instead of RI. On average of all feed samples, the potential GP was 9 mL/200 mg dry matter (DM) lower and the rate of GP was 3.1%/h lower with FI than RI. The estimation of kinetic data for FI included a lag phase of 1.51 h on average, whereas no lag phase was estimated for RI. Estimated parameters indicated an overall lower fermentation activity of FI than RI. However, there was a very similar progression of GP curves of RI-GP and FI-GP. RI-GP after 24 h was significantly correlated with FI-GP in the time period of 24 to 60 h (r = 0.973 – 0.982, including all feed samples). Linear regression analysis showed that RI-GP after 24 h can be estimated from FI-GP after 24 h (Slope = 1.02, R2 = 0.97) and 48 h (Slope = 1.1, R2 = 0.97) including the complete data set. Splitting the data set into roughages and concentrate feeds did not lead to distinctly higher estimation accuracy. In conclusion, sheep faeces can be used as an alternative inoculum for studying a wide range of different-quality feedstuffs. The standard gas test may be modified without the need for rumen-cannulated animals.

법적 판단에서의 증거평가 차원: 다차원 척도법 및 펴기의 적용

The purpose of this study was to explore the dimensions that laypersons use to evaluate evidence presented in criminal cases and to determine whether the dimensions differ according to legal decision using rankings of the importance of evidence. Multidimensional scaling was performed using the proximity matrix generated from the entire data set, and it was found that laypeople are likely to evaluate evidence according to the legal party from which it was presented (Dimension 1), while Dimension 2 may reflect the sequence of events or the objectivity of the evidence itself. The multidimensional scaling configurations for the guilty and not guilty decision were similar to the configuration of the total data, but the clustering of the evidence and the meaning of the dimensions tended to be clearer in the guilty decision configuration. When comparing vector model configurations of multidimensional unfolding using ranking data, no evidence or dimension was observed to be particularly important to those who were pro-defendant, but a majority of those who were pro-prosecution people valued the defense's evidence. Although this study is exploratory and has limited generalizability to laypersons' evidence evaluation dimensions, it is expected to provide essential knowledge for systemic understanding of the jury decision-making process.

Searching the InSight Seismic Data for Mars’s Background-Free Oscillations

Abstract Mars’s atmosphere has theoretically been predicted to be strong enough to continuously excite Mars’s background-free oscillations, potentially providing an independent means of verifying radial seismic body-wave models of Mars determined from marsquakes and meteorite impacts recorded during the Interior Exploration using Seismic Investigations, Geodesy, and Heat Transport (InSight) mission. To extract the background-free oscillations, we processed and analyzed the continuous seismic data, consisting of 966 Sols (a Sol is equivalent to a Martian day), collected by the Mars InSight mission using both automated and manual deglitching schemes to remove nonseismic disturbances. We then computed 1-Sol-long autocorrelations for the entire data set and stacked these to enhance any normal-mode peaks present in the spectrum. We find that while peaks in the stacked spectrum in the 2–4 mHz frequency band align with predictions based on seismic body-wave models and appear to be consistent across the different processing and stacking methods applied, unambiguous detection of atmosphere-induced free oscillations in the Martian seismic data nevertheless remains difficult. This possibly relates to the limited number of Sols of data that stack coherently and the continued presence of glitch-related signal that affects the seismic data across the normal-mode frequency range (∼1–10 mHz). Improved deglitching schemes may allow for clearer detection and identification in the future.

Fake Review Detection Using Machine learning and Deep Learning

The widespread adoption of Web 2.0 platforms has enabled consumers to share their opinions on products and services, influencing purchasing decisions. However, the proliferation of spam reviews has undermined the credibility of online reviews. This study aims to identify and evaluate existing approaches for detecting reviews, individual spammers, and Organizations. We categorized machine learning (ML) and deep learning (DL) techniques used for Review detection and assessed their effectiveness. Our findings indicate that accuracy is the most frequently used metric (25%), followed by recall (24%) and precision (22%). Additionally, we observed that utilizing the entire Amazon dataset can enhance the performance of F-measure, AUC, and F1-score metrics by 7%. Our study concludes that SMS spam filtering strategies are often effective in combating spam reviews. Furthermore, we developed a taxonomy of existing methodologies and observed a significant number of studies employing SMS anti-spam applications. This research uncovered innovative applications of ML and DL to spam review detection, offering a novel approach to addressing this issue. Our findings provide both academics and practitioners with a deeper understanding of the challenges in spam review identification and potential avenues for improvement using ML techniques.

BiodiViz: Leveraging NER and RE for Automated Knowledge Graph Generation in Biodiversity Research

In biodiversity research, the integration of machine learning and data visualization is increasingly important for uncovering valuable insights from academic literature. This study introduces an innovative knowledge graph application, BiodiViz, designed to translate intricate text into intuitive visual representations, fostering a deeper comprehension of biodiversity relationships. BiodiViz uses the top-performing Named Entity Recognition (NER) and Relation Extraction (RE) models to automatically generate a comprehensive knowledge graph for biodiversity research. The NER model extracts and categorizes entities like organisms, phenomena, and habitats, while the RE model identifies relationships such as "have," "occur in," and "influence" from the BiodivNERE dataset (Abdelmageed et al. 2022). These entities and relationships are organized into nodes and edges within a graph. Researchers input text into BiodiViz, producing a visual knowledge graph that simplifies the analysis of complex biodiversity data, reducing manual effort and enhancing efficiency. Named Entity Recognition & Relation Extraction BiodiViz leverages advanced Bidirectional Encoder Representations from Transformers (BERT)-based Large Language Models (LLMs) (Rogers et al. 2020), fine-tuned specifically for NER and RE tasks using the BiodivNERE dataset. The fine-tuning process involved various models, including BERT (Devlin et al. 2019), ELECTRA (Clark et al. 2020), and BiodivBERT (Abdelmageed et al. 2023). These models were evaluated for performance using the results of their F1-score as the main metric, which is the harmonic mean of precision (the proportion of true positive results among all positive predictions) and recall (the proportion of true positive results among all actual positives), with BiodivBERT achieving an F1-score of 77.16% for the NER task, while BERT excelled in the RE task with an F1-score of 81.28%. Rigorous hyperparameter optimization further enhanced the performance of BiodivBERT in the RE task by 3.38%. The BiodivNERE corpora by Abdelmageed et al. (2022) were used to fine-tune several models for NER and RE tasks in the biodiversity domain. The first corpus from the BiodivNERE corpora is BiodivNER, which is a gold standard dataset (manually labelled test corpora) for evaluating NER tasks. The fine-tuning process employed the token classification method from the Hugging Face library (Hugging Face 2023b), which assigns labels to each token in a sequence. Experiments were conducted with a batch size of four, meaning the model processes four examples/rows of data at a time before making an update to improve its learning. This is due to the constraints of the NVIDIA® GeForce RTX™ 3060 graphics processor. (NVIDIA 2024) Model performance was evaluated using the seqeval library (Nakayama 2018), focusing on accuracy, precision, recall, and F1 scores. For text classification, the second corpus, BiodivRE, was utilized, following previous research recommendations to explore fine-tuning settings for BiodivBERT. Hyperparameter optimization (Feurer and Hutter 2019) was conducted using Hugging Face’s Trainer API with an Optuna backend (Hugging Face 2023a), concentrating on learning rate and the number of training epochs (i.e., the number of complete passes through the entire dataset during model training). The BiodiViz Knowledge Graph Application The fine-tuned NER and RE models with the best F1-scores—BiodivBERT and BERT, respectively—were integrated into the knowledge graph application. Fig. 1 illustrates the flowchart of the application pipeline. Each sentence in the input text will go through the NER model to identify and label the entities within the sentence. Subsequently, these labeled entities, together with the original sentence, will be input into the RE model. The RE model will analyze every pair of entities for a potential relation and output the type of relation they share. The application will then utilize this data to create a graph with appropriate labels and color-coding. An example of the application's user interface with the knowledge graph is shown in Fig. 2. This study highlights the practical application of machine learning and data visualization in advancing biodiversity research, emphasizing the importance of developing user-friendly tools to support scientific exploration and discovery. The BiodiViz application, including the code and resources, is available on GitHub*1, providing an accessible tool for biodiversity researchers to streamline their analyses.

Deep Learning-Based Microscopic Damage Assessment of Fiber-Reinforced Polymer Composites.

Fiber-reinforced polymers (FRPs) are increasingly being used as substitutes for traditional metallic materials across various industries due to their exceptional strength-to-weight ratio. However, their orthotropic properties make them prone to multiple forms of damage, posing significant challenges in their design and application. During the design process, FRPs are subjected to various loading conditions to study their microscopic damage behavior, typically assessed through scanning electron microscopy (SEM). While SEM provides detailed insights into fracture surfaces, the manual analysis of these images is labor-intensive, time-consuming, and subject to variability based on the observer's expertise. To address these limitations, this research proposes a deep learning-based approach for the autonomous microscopic damage assessment of FRPs. Several computationally efficient pre-trained deep learning models, such as DenseNet121, NasNet Mobile, EfficientNet, and MobileNet, were evaluated for their performance in identifying different damage modes autonomously, thus reducing the need for manual interpretation. SEM images of FRPs with five distinct failure modes were used to validate the proposed method. These failure modes include three fiber-based failures such as fiber breakage, fiber pullout, and mixed-mode failure, and two matrix-based failures such as matrix brittle failure and matrix ductile failure. The entire dataset is divided into train, validation, and test sets. Deep learning models were established by training on train and validation sets for five failure modes, while the test set was used as the unseen data to validate the models. The models were assessed using various evaluation metrics on an unseen test dataset. Results indicate that the EfficientNet model achieved the highest accuracy of 97.75% in classifying the failure modes. The findings demonstrate the effectiveness of employing deep learning techniques for microscopic damage assessment, offering a more efficient, consistent, and scalable solution compared to traditional manual analysis.

Dependence of the polytropic behaviour of solar wind protons on temperature anisotropy and plasma beta near L1

A polytropic process describes the transition of a fluid from one state to another through a specific relationship between the fluid density and temperature, while the value of the polytropic index that governs this relationship determines the heat transfer and the effective degrees of freedom of that specific process. In this paper we investigate in depth the relationship between the proton effective polytropic index gamma in the solar wind, the proton anisotropy alpha , and plasma beta , while---for the first time to our knowledge to such an extent---we further investigate the dependence of the partial (with respect to the magnetic field) polytropic index to both the above-mentioned plasma parameters. To this end we use the entire Wind dataset spanning the 1995 to 2023 time period to derive the distributions of the polytropic index in the near-Earth space (L1). Our results indicate that the proton gamma increases with increasing proton anisotropy and decreases with increasing plasma beta . Finally, we show that even though the average (over long time periods) total and partial proton polytropic index values are very close, these values correspond to isotropic plasma alone, with a further balance between the thermal and magnetic pressure.On the contrary, for shorter time periods and/or specific solar wind structures, where the proton anisotropy and plasma beta exhibit deviations from these average values, the partial proton polytropic index exhibits significant variation that is dependent on the anisotropy and on plasma beta .