西北民族大学

On December 18, 2023, a 6.2-magnitude earthquake in Jishishan, Gansu Province triggered a large-scale loess liquefaction flow-slip disaster in Zhongchuan Township, resulting in 20 fatalities. It is crucial to investigate the liquefaction characteristics and occurrence mechanism of this disaster to understand the long-distance and low-angle liquefaction instability slip mechanism of loess slopes. Therefore, field investigations were conducted at the Zhongchuan liquefied flow-slip site, and undisturbed samples of saturated loess were collected from the site. Based on indoor dynamic triaxial tests, scanning electron microscopy (SEM), and particle and crack analysis system (PCAS) analysis, this study systematically investigated the cumulative evolution of dynamic pore water pressure and dynamic strain, stress path response, and mesostructural change during liquefaction, and further explored the underlying mechanism of the earthquake liquefaction disaster. The results demonstrated that the dynamic strain and dynamic pore water pressure accumulated with the number of vibrations, transitioning from the initial steady slow growth to rapid and fluctuating growth. During liquefaction, the area of the stress–strain hysteresis loop increased with an increase in the number of vibrations, the slope of its long axis decreased, and the short axis elongated. The effective stress path exhibited a linear variation under different dynamic stresses, and the average effective stress in cyclic shear continued to decrease but rarely reached the state of initial liquefaction. The undisturbed loess was characterized by a point-contact-dominated and weakly cemented overhead macropore structure with clear particle boundaries. After liquefaction, the soil structure was damaged, the number of pores increased, and the shape coefficient distribution indicated that the proportion of irregular and slender pores increased significantly. Under the coupling effect of the seismic load and hydraulic force, the soil skeleton deforms rapidly, and the failure of cementation and the contact system leads to the collapse of the overhead structure, redistribution of particle migration, pore collapse, and pore connectivity. These processes caused a surge in dynamic pore water pressure and a sharp decrease in effective stress, ultimately inducing liquefaction.

To explore the microbial community dynamics,differential metabolites and potential metabolic pathways in the milk of yak and cattle-yak in Gannan area,this study selected 6 yaks and 6 cattle-yak from the same natural grazing condition.Milk samples were collected from these animals for analysis of microbial communities and metabolites by single-molecule real-time sequencing and non-targeted metabolomics.The results showed that there was no significant difference in the microbial diversity or community composition between yak and cattle-yak milk(P>0.05),while the bacterial abundance in the cattle-yak milk was higher than that in yak milk.The dominant bacterial phyla in the two types of milk were Proteobacteria,Firmicutes and Bacteroidota.The dominant genera in yak milk were Bacteroides,unclassified_Sphingobacteriaceae,unclassified_Muribaculaceae,Lachnospiraceae_NK4A136_group,and Prevotellaceae_UCG_001,while those in cattle-yak milk were Lactobacillus,Bacteroides,and Streptococcus.The metabolites D-glucosamine-6-phosphate,indole-3-methanol,N-methyl-L-aspartic acid,histidine betaine,and histidine-glutamate in yak milk might enhance its nutritional value by improving lipid stability,antioxidant capacity,and the efficiency of protein synthesis;while the metabolites linoleic acid and 3-methoxy-4-hydroxybenzaldehyde in cattle-yak milk might affect its oxidative stability and flavor characteristics.In conclusion,under the same grazing conditions,there was no significant difference in microbial diversity between the milk of yak and cattle-yak.However,the microbial communities of cattle-yak milk exhibited higher richness than those of yak milk.The metabolic characteristics of yak milk were related to nutritional fortification and antioxidant/protein synthesis,while the metabolic characteristics of cattle-yak milk might be related to the regulation of oxidative stability and flavor characteristics.

Ultrasound imaging devices enable dynamic visualization and recording of articulatory physiology,specifically the tongue,during speech production.The processing and analysis of ultrasound tongue imaging(UTI)data hold signifi-cant importance for practical linguistics,experimental phonetic research,and speech engineering applications.Within these domains,the dynamic analysis of tongue movement trajectories and the quantitative analysis of lingual posture based on tongue contour extraction constitute critical technical challenges in practical implementation.However,UTI analysis faces substantial difficulties due to inherent noise characteristics,such as image blurring and indistinct tongue boundaries.These challenges manifest primarily as obstacles in accurately extracting key lingual positional features and reliably track-ing dynamic tongue contours throughout speech sequences.Deep learning(DL)techniques,leveraging their powerful capacity for hierarchical feature learning and representation,have demonstrated considerable success in advancing UTI pro-cessing and analysis.Significant progress has been achieved in several core areas,including the automated extraction of tongue contours from noisy ultrasound frames,the construction of sophisticated models representing complex tongue motion dynamics,and the facilitation of deeper integration into advanced speech engineering applications.This comprehensive overview details the fundamental principles of medical ultrasound imaging and its specialized adaptation for lingual visual-ization(i.e.,UTI).It systematically examines the persistent challenges encountered in UTI data processing.Further-more,it concisely outlines the methodological foundations of DL,encompassing key architectures like convolutional neural networks,recurrent neural networks,and their variants(e.g.,long short-term memory networks),which are particularly suited for spatial and spatiotemporal data analyses.The core of this review is that it synthesizes and critically evaluates the modeling mechanisms and specific applications of DL techniques across various tasks within UTI processing and analysis.Key application areas include,but are not limited to the following:ultrasound-video synchronization:precise temporal alignment of ultrasound image sequences with concurrently recorded video(e.g.,lip movement)and acoustic signals for multimodal analysis.Biometric identification:exploiting unique individual tongue movement patterns derived from UTI for speaker recognition or verification systems.Multimodal learning:integrating UTI data with complementary modalities such as audio signals,electromagnetic articulography,or visual speech information(video)to build robust and comprehensive models of speech production using fusion techniques.Silent speech interfaces(SSIs):enabling speech recognition or syn-thesis solely from articulatory movements captured by UTI,which is particularly valuable in noisy environments or for indi-viduals with voice impairments.Acoustic-articulatory inversion modeling:predicting the underlying articulatory configura-tions(represented by UTI data)from the acoustic speech signal or synthesizing speech from articulatory data.Language learning and speech rehabilitation:providing visual biofeedback of tongue positioning and movement to second language learners aiming to master new phonemes or to individuals undergoing therapy for speech sound disorders(e.g.,apraxia and dysarthria).Innovative interaction and artistic expression:exploring novel human-computer interaction paradigms and artistic performances utilizing real-time tongue movement tracking.For each application domain,this review systematically delineates the underlying theoretical frameworks,specific technical methodologies employed(detailing network architec-tures,input representations,loss functions,and training protocols),and established performance evaluation metrics used to assess the efficacy of the proposed DL solutions.Common metrics include contour extraction accuracy(e.g.,Dice coef-ficient and Hausdorff distance),tracking error,recognition accuracy,synthesis quality(e.g.,Mel cepstral distortion),and inference speed.Finally,this review critically examines the current state of the field and outlines compelling research prospects and future trends for DL in UTI processing and analysis.These aspects include the following:development of more efficient and lightweight models:enabling real-time processing on mobile or embedded devices for practical applica-tions like wearable SSI or biofeedback tools.Enhanced robustness and generalization:creating models that perform reliably across diverse speakers,accents,imaging devices,and recording conditions,potentially through advanced domain adapta-tion,data augmentation,or self-supervised/semisupervised learning paradigms.Explainable artificial intelligence(AI)for UTI:moving beyond black-box predictions to develop interpretable DL models that provide insights into how articulatory features are learned and utilized,fostering trust and deep linguistic understanding.Integration with advanced speech syn-thesis(vocoding):combining high-quality articulatory-to-acoustic inversion models with state-of-the-art neural vocoders for natural-sounding speech synthesis from UTI.Large-scale,multispeaker UTI datasets:facilitating the training of generaliz-able and powerful models through collaborative efforts to create and share comprehensive,annotated datasets.Exploration of self-supervised and unsupervised learning:reducing the heavy reliance on manually annotated UTI data by leveraging the inherent structure within unlabeled ultrasound sequences.Cross-lingual and low-resource adaptation:developing tech-niques to effectively apply models trained on data-rich languages to under-resourced languages with limited UTI data.Com-bining UTI with other biosignals:investigating fusion with neural signals(e.g.,EEG)or other physiological data for next-generation brain-computer interfaces or advanced speech rehabilitation monitoring.Future progress hinges on sustained interdisciplinary collaboration between linguists,speech scientists,computer vision researchers,machine learning experts,and engineers.Beyond technical advancements,the ethical deployment and accessibility of UTI-DL systems war-rant significant attention.Ensuring data privacy in biometric applications,addressing algorithmic bias across diverse popu-lations,and developing cost-effective solutions for clinical and educational settings are crucial for responsible innovation.Furthermore,establishing standardized evaluation benchmarks and open-source frameworks will accelerate reproducibility and community-driven progress.The continued integration and innovation in computer science,particularly in AI and machine learning,alongside advancements in ultrasound imaging technology,are anticipated to significantly propel scien-tific research and disciplinary development within experimental phonetics and linguistics.DL stands as a pivotal catalyst,unlocking the full potential of UTI as a powerful tool for understanding speech production,developing novel technologies,and enhancing human communication and well-being.

With the widespread application of large language models(LLMs),their potential ethical risks and social impacts have increasingly attracted attention.In order to solve the problem of scarcity of Chinese values alignment datasets,this paper constructs the Values Alignment Dataset(VAD),a Chinese dataset designed for instruction fine-tuning of large language models.The dataset sources include three types:manually constructed instruction data,synthetic data generated through data augmentation techniques,and optimized translations of open-source data.The manually constructed instruction data is designed based on topics aligned with the ethics,morals,and value principles of Chinese society and rigorously reviewed.The synthetic corpus draws on the Evol-Instruct method and is generated by depth and breadth evolution based on the CValues dataset,and the translated corpus is based on the SafeRLHF dataset,where high-quality Q&A pairs are selected for translation and optimization.In terms of quality control,the dataset employs the Instruction-Following Difficulty(IFD)filtering method.Samples with low IFD scores were removed by comparing the loss difference between the model's response generation with and without instruction context.Subsequently,manual review ensured that the data complied with the values corpus standard and the Chinese national context.Finally,29,000 high-quality samples were screened,stored in JSON format and contained five attributes:instruction,input,output,source,and categorization.The classification spans three levels:person,social,and nation.The VAD dataset has significant application value in areas such as ethics,socially sensitive issues,and cultural diversity.It supports the fine-tuning of LLMs to achieve values alignment in complex social contexts,reduces the generation of harmful content,and contributes to advancing the safety research and practical applications of Chinese large language models.

Text detection is a significant research area in natural language processing,aiming to identify the source and authenticity of text content to help users access more credible information amidst the vast amount of online data.To address the scarcity of text detection datasets,this study constructs a Multi-Domain Text Detection Dataset(MDD-TD).The dataset comprises translation-optimized open-source data,web data collected through web crawling,and synthetic data generated using prompt-augmentation strategies.Translation data are obtained by translating and refining the question-answer corpus from the SimpleAI/HC3 dataset.Web-sourced data are collected from the Weibo and Douban platforms through web crawling.Synthetic data are generated following a rule-driven approach,leveraging existing translation and web-sourced data with multiple prompt enhancement strategies.To ensure data quality,this study employs the PPL method to remove texts with abnormal perplexity values.Semantic similarity-based deduplication is further applied further used,combined with manual review to conduct secondary filtering of low-quality samples.Ultimately,31,721 high-quality samples are retained and stored in JSON format.Each sample contains three labels:"text","source",and"label",where"text"represents the text to be detected,"source"indicates the data origin,and"label"denotes the authenticity of the text.Evaluation experiments are conducted using two publicly available detection models,achieving an accuracy of 97.13％in the review domain for source detection and 98.87％in the news domain for authenticity detection.These results demonstrate that the dataset can serve as an effective resource for training and evaluating large language models in text detection tasks.