Matrix Factorization Method Research Articles

Prediction of miRNA-disease associations by neural network-based deep matrix factorization.

MicroRNA(miRNA) is a class of short non-coding RNAs with a length of about 22 nucleotides, which participates in various biological processes of cells. A number of studies have shown that miRNAs are closely related to the occurrence of cancer and various human diseases. Therefore, studying miRNA-disease associations is helpful to understand the pathogenesis of diseases as well as the prevention, diagnosis, treatment and prognosis of diseases. Traditional biological experimental methods for studying miRNA-disease associations have disadvantages such as expensive equipment, time-consuming and labor-intensive. With the rapid development of bioinformatics, more and more researchers are committed to developing effective computational methods to predict miRNA-disease associations in roder to reduce the time and money cost of experiments. In this study, we proposed a neural network-based deep matrix factorization method named NNDMF to predict miRNA-disease associations. To address the problem that traditional matrix factorization methods can only extract linear features, NNDMF used neural network to perform deep matrix factorization to extract nonlinear features, which makes up for the shortcomings of traditional matrix factorization methods. We compared NNDMF with four previous classical prediction models (IMCMDA, GRMDA, SACMDA and ICFMDA) in global LOOCV and local LOOCV, respectively. The AUCs achieved by NNDMF in two cross-validation methods were 0.9340 and 0.8763, respectively. Furthermore, we conducted case studies on three important human diseases (lymphoma, colorectal cancer and lung cancer) to validate the effectiveness of NNDMF. In conclusion, NNDMF could effectively predict the potential miRNA-disease associations.

Clinical Genetic Features and Neoadjuvant Chemotherapy Response in HER2-Low Breast Cancers: A Retrospective, Multicenter Cohort Study.

The objective of this work is to reveal differences in clinical and genetic features, as well as neoadjuvant chemotherapy (NAC) response, between HER2-low and HER2-zero or HER2-positive breast cancers. A total of 245 female patients with breast cancer were retrospectively enrolled from seven hospitals. Core needle biopsy (CNB) samples were collected before NAC and used for next-generation sequencing by a commercial gene panel. Clinical and genetic features, as well as NAC response, were compared between HER2-low and HER2-zero or HER2-positive breast cancers. The nonnegative matrix factorization (NMF) method was applied to cluster the C-Score of enrolled cases to reveal the intrinsic features of each HER2 subgroup. A total of 68 (27.8%) cases are HER2-positive, 117 (47.8%) cases are HER2-low, and 60 (24.5%) cases are HER2-zero. HER2-low breast cancers have a significantly lower pathologic complete response (pCR) rate than HER2-positive and HER2-zero breast cancers (p < 0.050 for all comparisons). Compared with HER2-low breast cancers, HER2-positive cases have higher rates of TP53 mutation, TOP2A amplification, and ERBB2 amplification, as well as lower rates of MAP2K4 mutation, ESR1 amplification, FGFR1 amplification, and MAPK pathway alteration (p < 0.050 for all comparisons). After clustering HER2-low cases by the NMF method, 56/117 (47.9%) are in cluster 1, 51/117 (43.6%) are in cluster 2, and 10/117 (8.5%) are in cluster 3. HER2-low cases in cluster 2 have the lowest pCR rate among the three clusters (p < 0.050). HER2-low breast cancers have significant genetic differences from HER2-positive cases. Genetic heterogeneity exists in HER2-low breast cancers and impacts on NAC response in this subgroup.

Adaptive Kernel Graph Nonnegative Matrix Factorization

Nonnegative matrix factorization (NMF) is an efficient method for feature learning in the field of machine learning and data mining. To investigate the nonlinear characteristics of datasets, kernel-method-based NMF (KNMF) and its graph-regularized extensions have received much attention from various researchers due to their promising performance. However, the graph similarity matrix of the existing methods is often predefined in the original space of data and kept unchanged during the matrix-factorization procedure, which leads to non-optimal graphs. To address these problems, we propose a kernel-graph-learning-based, nonlinear, nonnegative matrix-factorization method in this paper, termed adaptive kernel graph nonnegative matrix factorization (AKGNMF). In order to automatically capture the manifold structure of the data on the nonlinear feature space, AKGNMF learned an adaptive similarity graph. We formulated a unified objective function, in which global similarity graph learning is optimized jointly with the matrix decomposition process. A local graph Laplacian is further imposed on the learned feature subspace representation. The proposed method relies on both the factorization that respects geometric structure and the mapped high-dimensional subspace feature representations. In addition, an efficient iterative solution was derived to update all variables in the resultant objective problem in turn. Experiments on the synthetic dataset visually demonstrate the ability of AKGNMF to separate the nonlinear dataset with high clustering accuracy. Experiments on real-world datasets verified the effectiveness of AKGNMF in three aspects, including clustering performance, parameter sensitivity and convergence. Comprehensive experimental findings indicate that, compared with various classic methods and the state-of-the-art methods, the proposed AKGNMF algorithm demonstrated effectiveness and superiority.

Analysis of cerebral palsy gait based on movement primitives

BackgroundCerebral palsy is the most prevalent motor disorder among children. Despite extensive studies on motor modularity of gait of children with cerebral palsy, kinematic modularity of their gait has not been addressed which is the main goal of this study. MethodsThe kinematics of the gait of 13 typical development children and 188 children with cerebral palsy was captured and analyzed, where the cerebral palsy children were grouped into True, Jump, Apparent, and Crouch. Non-negative matrix factorization method was used to extract the kinematic modulus of each group, which were then clustered to find their characteristic movement primitives. The movement primitives of groups were then matched based on the similarity of their activation profiles. FindingsThe number of movement primitives was three for the Crouch group, four for the other cerebral palsy groups, and five for the typical development group. Compared to the typical development children, the kinematic modules and activations of the cerebral palsy groups involved higher variability and co-activation, respectively (P < 0.05). Three temporally matched movement primitives were shared by all groups, but with altered structures. InterpretationThe gait of children with cerebral palsy involved lower complexity and higher variability due to the reduced and inconsistent kinematic modularity. Three basic movement primitives were sufficient to prodcue the overall gait kinematics, as observed in the Crouch group. Other movement primitives, were responsible for providing smooth transitions between basic movement primitives, as seen in more complex gait patterns.

Community Detection in Multilayer Networks Based on Matrix Factorization and Spectral Embedding Method

Community detection remains a challenging research hotspot in network analysis. With the complexity of the network data structures increasing, multilayer networks, in which entities interact through multiple types of connections, prove to be effective in describing complex networks. The layers in a multilayer network may not share a common community structure. In this paper, we propose a joint method based on matrix factorization and spectral embedding to recover the groups not only for the layers but also for nodes. Specifically, the layers are grouped via the matrix factorization method with layer similarity-based regularization in the perspective of a mixture multilayer stochastic block model, and then the node communities within a layer group are revealed by clustering a combination of the spectral embedding derived from the adjacency matrices and the shared approximation matrix. Numerical studies show that the proposed method achieves competitive clustering results as the number of nodes and/or number of layers vary, together with different topologies of network layers. Additionally, we apply the proposed method on two real-world multilayer networks and obtain interesting findings which again highlight the effectiveness of our method.

Predicting Drug Synergy and Discovering New Drug Combinations Based on a Graph Autoencoder and Convolutional Neural Network.

Drug synergy is a crucial component in drug reuse since it solves the problem of sluggish drug development and the absence of corresponding drugs for several diseases. Predicting drug synergistic relationships can screen drug combinations in advance and reduce the waste of laboratory resources. In this research, we proposed a model that utilizes graph autoencoder and convolutional neural networks to predict drug synergy (GAECDS). Our methods include a graph convolutional neural network as an encoder to encode drug features and use a matrix factorization method as a decoder. Multilayer perceptron (MLP) was applied to process cell line features and combine them with drug features. Furthermore, the latent vectors generated during the encoding process are being used to predict drug synergistic scores using a convolutional neural network. By measuring prediction performance using AUC, AUPR, and F1 score, GAECDS superior to other state-of-the-art models. In addition, four pairs of the predicted top 10 drug combinations were found to work well enough for evaluation. The case study shows that the GAECDS approach is useful for identifying potential drug synergy.

Source apportionment and health impact assessment of atmospheric particulate matter in the city of São Carlos, Brazil

In this study, positive matrix factorization method was used for source apportionment of PM10 in the city of São Carlos from 2015 to 2018. The annual mean concentrations of PM10, 15 PAHs, 4 oxy-PAHs, 6 nitro-PAHs, 21 saccharides, and 17 ions in these samples were in the ranges 18.1 ± 6.99 to 25.0 ± 11.3 μg m−3 for PM10, 9.80 × 10−1 ± 2.06 to 2.03 ± 8.54 × 10−1 ng m−3 for ΣPAHs, 83.9 ± 35.7 to 683 ± 521 pg m−3 for Σoxy-PAHs, 1.79 × 10−2 ± 1.23 × 10−1 to 7.12 ± 4.90 ng m−3 for Σnitro-PAHs, 83.3 ± 44.7 to 142 ± 85.9 ng m−3 for Σsaccharides, and 3.80 ± 1.54 to 5.66 ± 4.52 μg m−3 for Σions. For most species, the concentrations were higher in the dry season than in the rainy. This was related not only to the low rainfall and relative humidity characteristic of the dry season but also to an increase in fire spots recorded in the region between April and September every year from 2015 to 2018. A 4-factor solution provided the best description of the dataset, with the four identified sources of PM10 being soil resuspension (28%), biogenic emissions (27%), biomass burning (27%), and vehicle exhaust together with secondary PM (18%). Although the PM10 concentrations were not above the limit established by local legislation, the epidemiological study showed that by reducing PM2.5 concentrations to the level recommended by the WHO, approximately 35 premature deaths per 100,000 population could be avoided annually. The results revealed that biomass burning continues to be one of the main anthropic sources of emissions to the atmosphere in the region, so it needs to be incorporated into the existing guidelines and policies to reduce the concentration of particulate matter to within the limits recommended by the WHO, in order to avoid premature deaths.

Parent and alkylated polycyclic aromatic hydrocarbon emissions from coal seam fire at Wuda, Inner Mongolia, China: characteristics, spatial distribution, sources, and health risk assessment.

The Wuda coalfield in Inner Mongolia is a vital coal base in China, and it is the hardest-hit area for coal fires (spontaneous combustion of coal seams and coal gangue). Using gas chromatography-mass spectrometry, this work tested the concentration and analyzed the characteristics, distribution, sources, and health risks of polycyclic aromatic compounds (PACs) in the surface soil of the Wuda District, including the coal mine, coal fire, agricultural, and background areas. The soil of coal mine and coal fire area were heavily polluted with PACs, with mean concentrations of 9107 and 3163µgkg-1, respectively, considerably higher than those in the agricultural (1232µgkg-1) and background areas (710µgkg-1). Alkyl polycyclic aromatic hydrocarbons (APAHs) were the dominant pollutants among these PACs, accounting for 60-81%. Alkyl naphthalenes and alkyl phenanthrenes are the primary pollutants in APAHs, accounting for 80-90% of the total amounts. Additionally, using the positive matrix factorization method, it can be concluded that the primary PAC sources are petrogenic sources, coal and biomass combustion, coal fires, and vehicle emissions. Finally, according to the cancer risk values of 16 PAHs, only the coal mine area showed a potential cancer risk. However, this result lacks a risk assessment of APAHs and underestimates the actual risk. The results of this study improved the understanding of PAC pollution in coal fire and surrounding areas and provided a reference for environmental and health risk investigations.

PM2.5-bound PAHs near a typical industrial park: Determining health risks associated with specific industrial sources

Source-specific potential health risk was proposed as an effective index for determining priority control sources of PM2.5-bound polycyclic aromatic hydrocarbons (PAHs) near an industrial park. However, industry-specific emission profiles were limited. In this study, 60 emission source samples from 13 industries and 139 p.m.2.5 samples near an industrial park were collected from 2020 to 2021. The PAH profiles of 12 industrial emission sources were constructed to address the difficulty of source apportionment refined to industrial categories. A diagnostic ratio method and a positive matrix factorization method (PMF) were used to qualitatively and quantitatively determine the emission sources of 16 p.m.2.5-bound PAHs. The incremental lifetime carcinogenic risk (ILCR) of PAH inhalation exposure for the local population was attributed to sources by a PMF-ILCR model. The results show that the local PAH pollution seriously exceeded (86.3% sampling days) the standard (GB3095-2012). The concentration of PAHs was 33.6 ± 32.6 ng/m3 and showed a seasonal trend of “heating season > autumn > spring”. There were five main sources of PM2.5-bound PAHs near the industrial park: heavy oil combustion (29.8%), coal and natural gas combustion (28.0%), vehicle emissions (19.9%), steel industry emissions (19.4%) and tar volatilization (2.9%). There is potential carcinogenic risk to both children (3.1 × 10−6) and adults (1.2 × 10−5). The mass contribution of each source to ambient PAHs is not equivalent to its contribution to potential health risk. The risks of PM2.5-bound PAHs almost all correspond to vehicle emissions (46.3%), heavy oil combustion (25.9%) and steel industrial sources (24.9%).

Robust design of semi-automated clustering models for 4D-STEM datasets

Materials discovery and design require characterizing material structures at the nanometer and sub-nanometer scale. Four-Dimensional Scanning Transmission Electron Microscopy (4D-STEM) resolves the crystal structure of materials, but many 4D-STEM data analysis pipelines are not suited for the identification of anomalous and unexpected structures. This work introduces improvements to the iterative Non-Negative Matrix Factorization (NMF) method by implementing consensus clustering for ensemble learning. We evaluate the performance of models during parameter tuning and find that consensus clustering improves performance in all cases and is able to recover specific grains missed by the best performing model in the ensemble. The methods introduced in this work can be applied broadly to materials characterization datasets to aid in the design of new materials.

Deep Nonnegative Matrix Factorization Using a Variational Autoencoder With Application to Single-Cell RNA Sequencing Data.

Single-cell RNA sequencing is used to analyze the gene expression data of individual cells, thereby adding to existing knowledge of biological phenomena. Accordingly, this technology is widely used in numerous biomedical studies. Recently, the variational autoencoder has emerged and has been adopted for the analysis of single-cell data owing to its high capacity to manage large-scale data. Many different variants of the variational autoencoder have been applied, and have yielded superior results. However, because it is nonlinear, the model does not provide parameters that can be used to explain the underlying biological patterns. In this paper, we propose an interpretable nonnegative matrix factorization method that decomposes parameters into those shared across cells and those that are cell-specific. Effective nonlinear dimension reduction was achieved via a variational autoencoder applied to the cell-specific parameters. In addition to achieving nonlinear dimension reduction, our model could estimate the cell-type-specific gene expression. To improve the estimation accuracy, we introduced log-regularization, which reflects the single-cell property. Overall, our approach displayed excellent performance in a simulation study and in real data analyses, while maintaining good biological interpretability.

Molecular features and predictive models identify the most lethal subtype and a therapeutic target for osteosarcoma.

Osteosarcoma is the most common primary malignant bone tumor. The existing treatment regimens remained essentially unchanged over the past 30 years; hence the prognosis has plateaued at a poor level. Precise and personalized therapy is yet to be exploited. One discovery cohort (n=98) and two validation cohorts (n=53 & n=48) were collected from public data sources. We performed a non-negative matrix factorization (NMF) method on the discovery cohort to stratify osteosarcoma. Survival analysis and transcriptomic profiling characterized each subtype. Then, a drug target was screened based on subtypes' features and hazard ratios. We also used specific siRNAs and added a cholesterol pathway inhibitor to osteosarcoma cell lines (U2OS and Saos-2) to verify the target. Moreover, PermFIT and ProMS, two support vector machine (SVM) tools, and the least absolute shrinkage and selection operator (LASSO) method, were employed to establish predictive models. We herein divided osteosarcoma patients into four subtypes (S-I ~ S-IV). Patients of S- I were found probable to live longer. S-II was characterized by the highest immune infiltration. Cancer cells proliferated most in S-III. Notably, S-IV held the most unfavorable outcome and active cholesterol metabolism. SQLE, a rate-limiting enzyme for cholesterol biosynthesis, was identified as a potential drug target for S-IV patients. This finding was further validated in two external independent osteosarcoma cohorts. The function of SQLE to promote proliferation and migration was confirmed by cell phenotypic assays after the specific gene knockdown or addition of terbinafine, an inhibitor of SQLE. We further employed two machine learning tools based on SVM algorithms to develop a subtype diagnostic model and used the LASSO method to establish a 4-gene model for predicting prognosis. These two models were also verified in a validation cohort. The molecular classification enhanced our understanding of osteosarcoma; the novel predicting models served as robust prognostic biomarkers; the therapeutic target SQLE opened a new way for treatment. Our results served as valuable hints for future biological studies and clinical trials of osteosarcoma.

DRaW: prediction of COVID-19 antivirals by deep learning—an objection on using matrix factorization

BackgroundDue to the high resource consumption of introducing a new drug, drug repurposing plays an essential role in drug discovery. To do this, researchers examine the current drug-target interaction (DTI) to predict new interactions for the approved drugs. Matrix factorization methods have much attention and utilization in DTIs. However, they suffer from some drawbacks.MethodsWe explain why matrix factorization is not the best for DTI prediction. Then, we propose a deep learning model (DRaW) to predict DTIs without having input data leakage. We compare our model with several matrix factorization methods and a deep model on three COVID-19 datasets. In addition, to ensure the validation of DRaW, we evaluate it on benchmark datasets. Furthermore, as an external validation, we conduct a docking study on the COVID-19 recommended drugs.ResultsIn all cases, the results confirm that DRaW outperforms matrix factorization and deep models. The docking results approve the top-ranked recommended drugs for COVID-19.ConclusionsIn this paper, we show that it may not be the best choice to use matrix factorization in the DTI prediction. Matrix factorization methods suffer from some intrinsic issues, e.g., sparsity in the domain of bioinformatics applications and fixed-unchanged size of the matrix-related paradigm. Therefore, we propose an alternative method (DRaW) that uses feature vectors rather than matrix factorization and demonstrates better performance than other famous methods on three COVID-19 and four benchmark datasets.

3PNMF-MKL: A non-negative matrix factorization-based multiple kernel learning method for multi-modal data integration and its application to gene signature detection.

In this current era, biomedical big data handling is a challenging task. Interestingly, the integration of multi-modal data, followed by significant feature mining (gene signature detection), becomes a daunting task. Remembering this, here, we proposed a novel framework, namely, three-factor penalized, non-negative matrix factorization-based multiple kernel learning with soft margin hinge loss (3PNMF-MKL) for multi-modal data integration, followed by gene signature detection. In brief, limma, employing the empirical Bayes statistics, was initially applied to each individual molecular profile, and the statistically significant features were extracted, which was followed by the three-factor penalized non-negative matrix factorization method used for data/matrix fusion using the reduced feature sets. Multiple kernel learning models with soft margin hinge loss had been deployed to estimate average accuracy scores and the area under the curve (AUC). Gene modules had been identified by the consecutive analysis of average linkage clustering and dynamic tree cut. The best module containing the highest correlation was considered the potential gene signature. We utilized an acute myeloid leukemia cancer dataset from The Cancer Genome Atlas (TCGA) repository containing five molecular profiles. Our algorithm generated a 50-gene signature that achieved a high classification AUC score (viz., 0.827). We explored the functions of signature genes using pathway and Gene Ontology (GO) databases. Our method outperformed the state-of-the-art methods in terms of computing AUC. Furthermore, we included some comparative studies with other related methods to enhance the acceptability of our method. Finally, it can be notified that our algorithm can be applied to any multi-modal dataset for data integration, followed by gene module discovery.

3D geometrical total variation regularized low-rank matrix factorization for hyperspectral image denoising

The total variation (TV) regularization has been widely used in hyperspectral image (HSI) denoising owing to its powerful capabilities in terms of structure preservation. However, existing TV terms ignore the geometrical structures, such as edges and textures, in HSIs. To this end, we present a novel 3D geometrical total variation (3D GTV) regularizer to capture the local information in HSIs. 3D GTV can preserve the spatial and spectral information in 3D cubes effectively due to the introduction of geometrical structures. By incorporating this term, a 3D GTV regularized low-rank matrix factorization (3DGTVLR) method is proposed for the noise removal of noisy HSIs. In the proposed method, different norms are adopted to model the structures in 3D cubes. More specifically, the L1 norm is employed to enhance the structural information in the 3D cubes containing edges or textures, while the L2 norm is considered to ensure the information in flat 3D cubes. The local structures in HSIs are represented more compatibly by the combination of L1 and L2 norms. Then, we derive an efficient algorithm based on the alternating direction method of multipliers (ADMM) to solve the optimization problem. Experiments are performed on synthetic and real HSI datasets and the effectiveness of the proposed method is demonstrated when compared with the state-of-the-art methods. Our code is available at https://github.com/RSMagneto/3DGTVLR.

Muscle activation patterns and muscle synergies reflect different modes of coordination during upper extremity movement.

A core issue in motor control is how the central nervous system generates and selects the muscle activation patterns necessary to achieve a variety of behaviors and movements. Extensive studies have verified that it is the foundation to induce a complex movement by the modular combinations of several muscles with a synergetic relationship. However, a few studies focus on the synergetic similarity and dissimilarity among different types of movements, especially for the upper extremity movements. In this study, we introduced the non-negative matrix factorization (NMF) method to explore the muscle activation patterns and synergy structure under 6 types of movements, involving the hand open (HO), hand close (HC), wrist flexion (WF), wrist extension (WE), supination (SU), and pronation (PR). For this, we enrolled 10 healthy subjects to record the electromyography signal for NMF calculation. The results showed a highly modular similarity of the muscle synergy among subjects under the same movement. Furthermore, Spearman's correlation analysis indicated significant similarities among HO-WE, HO-SU, and WE-SU (p < 0.001). Additionally, we also found shared synergy and special synergy in activation patterns among different movements. This study confirmed the theory of modular structure in the central nervous system, which yields a stable synergetic pattern under the same movement. Our findings on muscle synergy will be of great significance to motor control and even to clinical assessment techniques.

Association of RNA-modification "writer" genes with prognosis and response to immunotherapy in patients with low-grade glioma.

RNA modification is a key regulatory mechanism involved in tumorigenesis, tumor progression, and the immune response. However, the potential role of RNA modification "writer" genes in the immune microenvironment of gliomas and their effect on the response to immunotherapy remains unclear. The purpose of this study was to evaluate the role of RNA modification "writer" gene in the prognosis and immunotherapy response of low-grade glioma (LGG). The consensus non-negative matrix factorization (CNMF) method was used to identify different RNA modification subtypes. We used a novel eigengene screening method, the variable neighborhood learning Harris Hawks optimizer (VNLHHO), to screen for eigengenes among the RNA modification subtypes. We constructed a principal components analysis score(PCA_score)-based prognostic prediction model and validated it using an independent cohort. We also analyzed the association between PCA_score and the immune and molecular features of LGG. The results suggested that LGG can be divided into two different RNA modification-based subtypes with distinct prognostic and molecular features. High PCA_score was significantly associated with a poor prognosis in LGG and was an independent prognostic factor. A nomogram containing PCA_score and clinical features was constructed, and it showed a significant predictive value. PCA_score was negatively correlated with tumor purity and the abundance of CD4+ T cells in LGG patients. LGG patients with high PCA_score had lower Tumor Immune Dysfunction and Exclusion scores and showed an immunotherapy response. In conclusion, we report a novel RNA modification-based prognostic model for LGG that lays the foundation for evaluating LGG prognosis and developing more effective therapeutic strategies for these tumors.

Towards an automatic approach to modelling the circumgalactic medium: new tools for mock making and fitting of metal profiles in large surveys

Abstract We present two new tools for studying and modelling metal absorption lines in the circumgalactic medium. The first tool, dubbed ‘NMF Profile Maker’ (NMF–PM), uses a non-negative matrix factorization (NMF) method and provides a robust means to generate large libraries of realistic metal absorption profiles. The method is trained and tested on 650 unsaturated metal absorbers in the redshift interval z = 0.9–4.2 with column densities in the range of 11.2 ≤ log (N/cm−2) ≤ 16.3, obtained from high-resolution (R &amp;gt; 4000) and high-signal-to-noise ratio (S/N ≥ 10) quasar spectroscopy. To avoid spurious features, we train on infinite S/N Voigt models of the observed line profiles derived using the code ‘Monte-Carlo Absorption Line Fitter’ (MC–ALF), a novel automatic Bayesian fitting code that is the second tool we present in this work. MC–ALF is a Monte-Carlo code based on nested sampling that, without the need for any prior guess or human intervention, can decompose metal lines into individual Voigt components. Both MC–ALF and NMF–PM are made publicly available to allow the community to produce large libraries of synthetic metal profiles and to reconstruct Voigt models of absorption lines in an automatic fashion. Both tools contribute to the scientific effort of simulating and analysing metal absorbers in very large spectroscopic surveys of quasars like the ongoing Dark Energy Spectroscopic Instrument, the 4-m Multi-Object Spectroscopic Telescope, and the WHT Enhanced Area Velocity Explorer surveys.

Tourism poverty alleviation hotspots in China: Topic evolution and sustainable development

AbstractUnderstanding the hotspots and patterns of tourism poverty alleviation is of great significance to stakeholders. This study aims to comprehensively and scientifically understand the research hotspots, characteristics, trends, and future research directions in the field of tourism poverty alleviation, provide decision support for tourism stakeholders, improve the level of management and service quality of tourism, balance the distribution of tourism resources and interests, and promote high‐quality and sustainable development of tourism. First, we propose a dynamic nonnegative matrix factorization (DNMF) method to analyze the tourism poverty alleviation literature from 2012 to 2021, extract hot topics and topic words. Then, we propose a topic evaluation criteria to evaluate the effect of the extracted topics. Third, we propose a topic intensity model to calculate the intensity of hot topics that can track the evolution trend of hot topics in the past 10 years, and explore the future development trends of the tourism poverty alleviation. Finally, we put forward the countermeasures and suggestions for sustainable tourism poverty alleviation under the guidance of sustainable development theory. The experimental results show that DNMF has a good effect in topic extraction. The evolution process based on topic intensity reflects the evolution trend of China's tourism poverty alleviation. This study provides a theoretical foundations and technical support for the sustainable development of China's tourism poverty alleviation.

Source apportionment of polychlorinated dibenzo-p-dioxins and dibenzofurans in the sediments of an urban estuary.

Polychlorinated dibenzo-p-dioxins and dibenzofurans in the sediments of aquatic systems are a persistent global problem that poses serious health risks. Identifying the sources of dioxins in natural water systems and the extent of their contributions to observed sediment concentrations is important from a health advisory and mitigation perspective. This paper proposes novel distribution-based qualitative and quantitative methods as source apportionment techniques and alternatives to conventional source attribution methods. Using sampled data, air, runoff, industrial effluent, and industrial paper and pulp wastes were identified as four distinct dioxin contributors to concentrations found in the sediments of the test bed region: the Houston Ship Channel-San Jacinto River-Galveston Bay (HSC-SJR-GB) estuarine system that also includes 2 Superfund sites with dioxin contamination. Two qualitative methods, the Kullback-Leibler divergence (K-L divergence) and the Bhattacharya measure (BM), and a quantitative method, the L2 norm, were used to investigate the spatial and temporal sourcing patterns of dioxins in the system sediments. The results indicated a global contribution from air and runoff sources across the estuarine system and over time with more localized impacts of the Superfund sites and the industrial sources. The results using the developed methodologies were compared with the output from the more conventional positive matrix factorization (PMF) method. Statistically significant correlations were observed among source contributions from the proposed methods and the PMF method, with Spearman's ρ ranging between - 0.596 to - 0.963 and 0.652 to 0.719, demonstrating the utility of the sourcing approaches used in the study. Additionally, the proposed methods were found to be rigorous in terms of elucidating spatial and temporal changes in the sourcing of dioxin to the estuary, indicating their suitability for use for other contaminants and other estuarine systems.