• 2024-2024close
• Open Access close
• Closed Access close
• Open Source close
• Publications close
• Research software close
• Other research products close
• Chinese close
• Digital Humanities and Cultural Her... close

In the context of globalization, the reconstruction of material spaces and the evolution of social relations have accelerated the loss of emotions in traditional villages. Intangible heritage can play an important role in the emotional maintenance of traditional villages as an emotional carrier for its residents. Previous studies have been less involved in the synergistic relationship between the interactive game of power subjects and the evolution of emotions in the practice of intangible heritage. Particularly, research on the evolution of the emotional exchange mode is insufficient. Taking the Dong minority chorus of Huangdu Village as an example, this study adopts qualitative research methods, such as semi-structured interviews and participant observation, to construct an analytical framework of "daily life practice-emotional exchange", and to deeply explore the evolution process and motivational mechanism of the emotional exchange mode in the daily life practice of traditional village residents. The study found that: 1) According to the changes in the subject and relationship, motive and mode, resource and situation, perception and experience of emotional exchange in the natural, livelihood, institutional, and spiritual dimensions, the daily life practice of Huangdu Village can be divided into four stages: primitive equilibrium, passive compromising, active resisting, and regenerating. 2) In the process of intangible cultural heritage practices, the manipulation of capital and the suppression of power have broken the original balance of Huangdu Village, and the division of power and status among subjects has squeezed the living spaces of local residents, forcing them to become involved in power struggles. They resist the control of the "other" by means of physical empowerment and restatement of the local subjectivity, and ultimately strike a balance of power within the village. In the daily practice of intangible cultural heritage, the mode of residents' emotional exchange changes from reciprocity to general negotiation and production modes. 3) Emotional exchanges in traditional villages are produced during power struggles between residents and other subjects. When power is balanced, residents master the discourse of intangible cultural heritage and produce positive emotions such as attachment and belonging. When residents are suppressed by power and capital, they gradually lose discourse and produce negative emotions, such as a sense of crisis and separation. 4) The evolution of emotional exchange modes in the daily practices of traditional village residents is systematic. The evolution of the outer system pushes the kernel system to adapt, and the driving, pulling, and supporting forces promote the synergistic evolution of "daily life―emotional exchange―intangible cultural heritage practice" in Huangdu village. The evolution of emotional exchange patterns during the practice of the Dong minority chorus in Huangdu Village was an inevitable process for reconstructing the cultural subjectivity of local residents in the context of tourism development. Exploring emotional exchange patterns at different stages of daily life practices can help understand the developmental law of traditional villages and provide useful references for its emotional governance and sustainable development.

To better predict the lateral displacements of diaphragm walls during deep excavation, a long short-term memory (LSTM) multi-step prediction model is developed in this paper based on the LSTM algorithm. First, the multi-output strategy of multi-step prediction model is discussed. Then, the construction method of the LSTM multi-step prediction model is introduced in detail, and the two hyperparameters, i.e., the space and time dimensions of the model input set, are explored to improve the prediction accuracy of the model. Finally, the errors between the predicted values and the field monitoring data are analyzed based on an excavation project buried in water-rich sandy strata. The analysis results of three typical monitoring points indicate that the LSTM prediction model is characterized by solid generalization ability, and the relevant algorithm is practically helpful for improving and optimizing deformation prediction methods of deep excavation.

Crossmodal image-text retrieval involves retrieving relevant images or texts based on a query condition from the opposite modality. Its primary challenge lies in precisely quantifying the similarity metric used for feature matching between the two distinct modalities, playing an important role in mitigating the visual-semantic disparities between the heterogeneous realms of visual and linguistic domains. It has extensive applications in domains such as e-commerce product search and medical image retrieval. Traditional retrieval paradigms depend on harnessing deep learning techniques for extracting feature representations from images and texts. Crossmodal image-text retrieval learns semantic feature representations of disparate modal data by harnessing the formidable feature–extraction ability, subsequently mapping them into a shared semantic space for semantic alignment. However, this approach primarily depends on superficial data correlations, lacking the capacity to reveal the latent causal relationships underpinning the data. Moreover, owing to the inherent “black-box” nature of deep learning, the interpretability of model predictions often eludes human comprehension. In addition, an undue reliance on training data distributions impairs the generalization performance of the model. Consequently, the existing methods suffer the challenge of representing high-level semantic insights while maintaining interpretability. Causal inference, which endeavors to ascertain the causal effect of specific phenomena by isolating confounding factors by means of intervention, presents a novel avenue for enhancing the generalization capability and interpretability of deep models. Recently, researchers have sought to combine visual and linguistic tasks with the principles of causal inference. Accordingly, we introduce causal inference and embeds consensus knowledge into the bedrock of deep learning, and a novel causal image-text retrieval methodology with embedded consensus knowledge is proposed. Specifically, causal intervention is introduced into the visual feature extraction module, replacing correlated relationships with causal counterparts to cultivate common causal visual features. These features are then fused with the primal visual features acquired through bottom-up attention, resulting in a definitive visual feature representation. This study adopts the potent textual feature extraction ability of bidirectional encoder representations from transformers to address the shortfall in textual feature representation. Shared consensus knowledge between the two modal data is entwined, allowing for consensus-level feature representation learning image-text features. Empirical validation on the dataset MS-COCO and crossdataset experiments on the dataset Flickr30k substantiate the capacity of the proposed method to consistently enhance recall and mean recall in bidirectional image-text retrieval tasks. In summary, this pioneering approach endeavors to bridge the gap between visual and textual representations by combining causal inference principles and shared consensus knowledge within the framework of deep learning, thereby promising enhanced generalization and interpretability.

Maritime Power has gradually increased as a national strategy. In this process, gross marine products continue to grow, and the marine industry has become the most fundamental and critical object. The spatial layout and industrial organization of maritime enterprises are fundamental related tasks. Domestic research can be divided into two main categories, based on the data used. One is to use economic and social statistical data, which have a large spatial scope but large granularity and cannot reflect the details of the industrial layout. The other is to use point-of-interest data, which are often not fully mined owing to the heavy workload of data processing. Therefore, there is little relevant content on departmental and urban comparisons in the existing research. Four representative cities-Dalian, Qingdao, Ningbo, and Xiamen-were selected as the research areas. According to the Industrial Classification for Ocean Industries and Their Related Activities, the research objects were identified as the marine core layer, marine support layer, and marine peripheral layer industries and further refined into eight subcategories. This study is based on the information of maritime enterprises registered for business registration, and uses Python to crawl geographic coordinates to improve the spatial information of enterprises. An innovative task is to identify the industry categories of enterprises. This task was performed using fastText, Convolutional Neural Networks, and Recurrent Neural Network. Thus, a spatial enterprise information database, including multiple marine industry departments, was established. Kernel density analysis, standard deviational ellipse analysis, buffer analysis, and other methods were used. Finally, by comparing the visualization results of the marine industrial spatial layout in the four cities, we delved into the marine industrial spatial differentiation patterns. In the experiment, machine learning models, such as artificial neural networks, exhibited high accuracy and recall when completing human recognition tasks, and these methods were effective. Empirical research on the spatial layout and industrial organization of maritime enterprises revealed the following: 1)From the perspective of spatial pattern, the overall pattern is a balanced pattern of "large dispersion and small agglomeration." By comparing the distribution and organization of different types of marine industries, we found that there is industry agglomeration in the location selection of enterprises, resulting in industry agglomeration characteristics. The land sea relationship is reflected in the high-density single peak or "coastal zone-city center" Multimodal distribution pattern. 2) From the perspective of spatial organization mode, industrial clusters have multilevel hierarchical characteristics corresponding to population size and administrative levels. In addition to single core structures, multi core structures generally exhibit a "primary-secondary dual core" or "primary core-multiple radial" pattern, with spatial connections between core intervals forming a multi node axis or network structure. 3) From the perspective of spatial matching relationships, the elliptical area is positively related to the urban area, the direction of the long axis is close to the coastal direction, and the industrial distribution has a clear matching relationship with the urban center, ports, and other transportation hubs, bay terrain, coastline, and other spatial elements.

《赫拉克勒斯：神话与传承》是一场跨学科的展览，其同时采用两条叙事线索来重述古代神话赫拉克勒斯的十二试炼，并探讨了自后文艺复兴时期至今赫拉克勒斯在科学、技术和艺术领域的影响与应用。 此次展览是周泽荣博物馆致力于“接受研究”系列展览中的第二场展览。第一场展览《动物之神：古典与分类》是关于荷马史诗《特洛伊战》和《奥德赛》。展览中介绍了林奈的生物分类和命名系统，突出了拉丁神话学家文本在名称应用中的作用，其往往没有考虑到被命名动物的物理属性。然而，对于使用‘赫拉克勒斯’ 这个名称的时候，最重要的是考虑到动物、地点或发明物的身体特征，以便将它们与赫拉克勒斯的特征联系起来。此次陈列品包括古代雅典和后文艺复兴时期的艺术作，以及在我们周围世界中应用了赫拉克勒斯及其同伴或对手的名称的动物、植物和物品。

This paper explores the application and development trends of artificial intelligence (AI) technology, particularly machine learning and natural language processing in the field of failure analysis. Failure analysis is a crucial method for ensuring the reliability and safety of equipment, and is widely used in aerospace, automotive manufacturing, electronic devices, and other fields. Traditional failure analysis methods often rely on expert experience, which is time-consuming and laborious. By integrating AI’s powerful data processing capabilities with traditional methods, the accuracy and efficiency of analysis have been significantly enhanced. In terms of failure mode diagnosis, AI can rapidly and accurately identify various fault modes and provide precise diagnostic results. In failure cause diagnosis, AI integrates data from multiple sources to uncover complex failure factors and potential causal relationships, improving diagnostic reliability. In failure prediction, machine learning can accurately forecast material lifespan and strength, reducing experimental time and costs. In failure prevention, AI offers new approaches to effectively reduce the risk of failure and lower product maintenance costs. The paper also looks forward the future development prospects of AI in failure analysis and highlights challenges and recommendations in the areas, such as data quality improvement, model optimization, interdisciplinary collaboration, and ethical and safety issues.

Over the past two decades, language modeling (LM) has emerged as a primary methodology for language understanding and generation. This technology has become a cornerstone within the field of natural language processing (NLP). At its core, LM is designed to train models to predict the probability of the next word or token, thereby generating natural and fluent language. The advent of large language models (LLMs), such as Bidirectional Encoder Representations from Transformers and GPT-3, marks a significant milestone in the evolution of LM. These LLMs have left a profound impact on the field of artificial intelligence (AI) while also paving the way for advancements in other domains. This progression underscores the power and efficacy of AI, illustrating how the landscape of AI research has been reshaped by the rapid advancement of LLMs. This paper provides a comprehensive review of the evolution of LLMs, focusing on the technical architecture, model scale, training methods, optimization techniques, and evaluation metrics. Language models have evolved significantly over time, starting from initial statistical language models, moving onto neural network-based models, and now embracing the era of advanced pre-trained language models. As the scale of these models has expanded, so has their performance in language understanding and generation. This has led to notable results across various sectors, including education, healthcare, finance, and industry. However, the application of LLMs also presents certain challenges, such as data quality, model generalization capabilities, and computational resources. This paper delves into these issues, providing an analysis of the strengths and limitations of LLMs. Furthermore, the rise of LLMs has sparked a series of ethical, privacy, and security concerns. For instance, LLMs may generate discriminatory, false, or misleading information, infringe on personal privacy, or even be exploited for malicious activities such as cyber-attacks. To tackle these issues, this paper explores relevant technical measures, such as model interpretability, privacy protection, and security assessments. Ultimately, the paper outlines potential future research trends of LLMs. With ongoing enhancements to model scale and efficiency, LLMs are expected to play an even greater role in multimodal processing and societal impact. For example, by integrating information from different modalities, such as images and sound, LLMs can better understand and generate language. Additionally, they can be employed for societal impact assessment, providing support for policy formulation and decision-making. By thoroughly analyzing the current state of research and potential future directions, this paper aims to offer researchers valuable insights and inspiration regarding LLMs, thereby fostering further advancement in the field.

In order to more accurately predict the lateral deformation of retaining structures caused by foundation pit excavation, this paper adopts support the vector machine model, traditional artificial neural network model, and two kinds of recurrent neural network models considering temporal inputs to establish a prediction model for the maximum lateral deformation of retaining structures in different foundation pits, and for the same foundation pit under different working conditions. The results show that the artificial neural network can update and predict the deformation of the retaining structure in real time based on the measured data of the project, which is helpful for timely planning of the next construction process of the project. In the prediction of lateral deformation of retaining structures under different working conditions, the cyclic neural network model considering temporal inputs is better than the traditional artificial neural network model.

BackgroundLuminescence dating technology has made significant advancements in determining the chronology of archaeological materials subjected to low firing temperatures. However, the luminescence dating of archaeological materials subjected to high firing temperatures remains challenging.PurposeThis study aims to explore the luminescence emission spectrum characteristics and luminescence properties of high-firing temperature quartz to verify the feasibility of thermoluminescence (TL) signals from different bands in luminescence dating.MethodsFirstly, the high-firing temperature (about 950 °C) quartz extracted from pottery unearthed at the Lingjiatan archaeological site was taken as a case study, spectral measurement platform was established using a Risø DA-20 luminescence dating instrument coupled with an Andor spectrometer and a charge-coupled device camera to analyze the luminescence spectral properties of archaeological quartz with high firing temperatures. Then, five filter combinations and two photomultiplier tubes (PMTs) were used to compare the TL and isothermal thermoluminescence (ITL) sensitivities of blue and red emissions. Kinetic parameters for Blue TL and Red TL were determined by deconvolving the glow curves with the general-order equation. Finally, exposure experiments were conducted on the Blue and Red TL using a solar simulator. The single aliquot regenerative dose (SAR) protocol was implemented to assess the applicability of the Blue TL-SAR, Blue ITL-SAR, Red TL-SAR, Red ITL-SAR, and optically stimulated luminescence (OSL)-SAR methods for dating archaeological quartz exposed to high temperatures during production or use.ConclusionsThe spectral analysis reveals that the archaeological quartz subjected to high firing temperature exhibits significant Red TL emissions at approximately 620 nm, which is correlated with the TL peak at 375 °C. This Red TL at 375 °C exhibits a marked insensitivity to light. The multi-wavelength TL, multiwavelength ITL, and conventional OSL dating results are consistent with the known radiocarbon age within the error range. This study demonstrates the potential feasibility of using luminescence signals of different wavelengths for chronological studies of archaeological materials subjected to high firing temperatures.