Search Results (39)

The aboveground biomass (AGB) of individual trees is a critical indicator for assessing urban forest productivity and carbon storage. In the context of global warming, it plays a pivotal role in understanding urban forest carbon sequestration and regulating the global carbon cycle. Recent advances in light detection and ranging (LiDAR) have enabled the detailed characterization of three-dimensional (3D) structures, significantly enhancing the accuracy of individual tree AGB estimation. This review examines studies that use LiDAR-derived 3D structural metrics to model and estimate individual tree AGB, identifying key metrics that influence estimation accuracy. A bibliometric analysis of 795 relevant articles from the Web of Science Core Collection was conducted using R Studio (version 4.4.1) and VOSviewer 1.6.20 software, followed by an in-depth review of 80 papers focused on urban forests, published after 2010 and selected from the first and second quartiles of the Chinese Academy of Sciences journal ranking. The results show the following: (1) Dalponte2016 and watershed are more widely used among 2D raster-based algorithms, and 3D point cloud-based segmentation algorithms offer greater potential for innovation; (2) tree height and crown volume are important 3D structural metrics for individual tree AGB estimation, and biomass indices that integrate these parameters can further improve accuracy and applicability; (3) machine learning algorithms such as Random Forest and deep learning consistently outperform parametric methods, delivering stable AGB estimates; (4) LiDAR data sources, point cloud density, and forest types are important factors that significantly affect the accuracy of individual tree AGB estimation. Future research should emphasize deep learning applications for improving point cloud segmentation and 3D structure extraction accuracy in complex forest environments. Additionally, optimizing multi-sensor data fusion strategies to address data matching and resolution differences will be crucial for developing more accurate and widely applicable AGB estimation models. Full article

This study examines whether firm managers strategically use tax avoidance to address climate risks, with a specific focus on strategies employed to reduce corporate income tax liabilities, and this study incorporates the moderating role of ESG performance and is ground in stakeholder theory to highlight the balance between sustainability and corporate profit expectations. Using the secondary data from Chinese A-listed companies during 2017–2023, the findings reveal that firms increasingly adopt tax avoidance practices in response to rising climate risks. More specifically, strong ESG performance positively moderates this relationship, underscoring its role in shaping socially and ethically responsible strategies to tackle sustainability challenges. By employing panel data analysis and addressing endogeneity through instrumental variable tests, Propensity Score Matching, and the Heckman test, this study provides robust results. These findings contribute to the literature on tax avoidance and provide practical insights for actionable ESG initiatives. For firms, these include improving transparency in tax reporting and integrating sustainability metrics into corporate ESG framework for firms. For tax authority, they involve upgrading the tax-related big data supervision system and fostering alignment between corporate practices and government policies. Full article

Background: The aim of this study was to explore the association between long-term exposure to particulate matter with an aerodynamic diameter <2.5 μm (PM_2.5) and its components and severe obesity in children and adolescents. Methods: Data for children and adolescents aged 9–18 in this cross-sectional study were obtained from the 2019 Chinese National Survey on Students’ Constitution and Health (CNSSCH). Data for PM_2.5 and its components were obtained from the Tracking Air Pollution in China (TAP) dataset and matched with information on these children. Logistic regression models were used to assess the risk of severe obesity associated with long-term exposure to PM_2.5 and its components. Results: A total of 160,205 children were included in the analysis. Long-term exposure to PM_2.5 may increase the odds of severe childhood obesity, with this effect being more pronounced in girls. Among boys, the component with the most significant impact on severe obesity was organic matter (OM). The impact of PM_2.5 and its components on severe obesity was greater in children from low economic and low parental education level households. Children with unhealthy lifestyle habits have higher odds of severe obesity due to long-term exposure to PM_2.5 and its components. Conclusions: The findings of this research support the development of strategies aimed at addressing severe obesity in children, suggesting that adopting healthy lifestyle practices could mitigate the odds of severe obesity due to PM_2.5 and its components. There is a need for an increased focus on children in economically underdeveloped areas and those with unhealthy lifestyle habits, particularly those in rural areas and those who do not engage in adequate physical activity or get enough sleep. Full article

The combination of UAV-LiDAR and LiDAR-SLAM (Simultaneous Localization and Mapping) technology can overcome the scanning limitations of different platforms and obtain comprehensive 3D structural information of forest stands. To address the challenges of the traditional registration algorithms, such as high initial value requirements and susceptibility to local optima, in this paper, we propose a high-precision, robust, NDT-VGICP registration method that integrates voxel features to register UAV-LiDAR and LiDAR-SLAM point clouds at the forest stand scale. First, the point clouds are voxelized, and their normal vectors and normal distribution models are computed, then the initial transformation matrix is quickly estimated based on the point pair distribution characteristics to achieve preliminary alignment. Second, high-dimensional feature weighting is introduced, and the iterative closest point (ICP) algorithm is used to optimize the distance between the matching point pairs, adjusting the transformation matrix to reduce the registration errors iteratively. Finally, the algorithm converges when the iterative conditions are met, yielding an optimal transformation matrix and achieving precise point cloud registration. The results show that the algorithm performs well in Chinese fir forest stands of different age groups (average RMSE—horizontal: 4.27 cm; vertical: 3.86 cm) and achieves high accuracy in single-tree crown vertex detection and tree height estimation (average F-score: 0.90; R² for tree height estimation: 0.88). This study demonstrates that the NDT-VGICP algorithm can effectively fuse and collaboratively apply multi-platform LiDAR data, providing a methodological reference for accurately quantifying individual tree parameters and efficiently monitoring 3D forest stand structures. Full article

Chinese dialogue generation faces multiple challenges, such as semantic understanding, information matching, and response fluency. Generative dialogue systems for Chinese conversation are somehow difficult to construct because of the flexible word order, the great impact of word replacement on semantics, and the complex implicit context. Existing methods still have limitations in addressing these issues. To tackle these problems, this paper proposes an improved Chinese dialogue generation model based on transformer architecture. The model uses a multi-layer transformer decoder as the backbone and introduces two key techniques, namely incorporating pre-trained language model word embeddings and optimizing the sparse Softmax loss function. For word-embedding fusion, we concatenate the word vectors from the pre-trained model with character-based embeddings to enhance the semantic information of word representations. The sparse Softmax optimization effectively mitigates the overfitting issue by introducing a sparsity regularization term. Experimental results on the Chinese short text conversation (STC) dataset demonstrate that our proposed model significantly outperforms the baseline models on automatic evaluation metrics, such as BLEU and Distinct, with an average improvement of 3.5 percentage points. Human evaluations also validate the superiority of our model in generating fluent and relevant responses. This work provides new insights and solutions for building more intelligent and human-like Chinese dialogue systems. Full article

Hearing loss is a frequently observed complication of type 2 diabetes (T2D). Emerging evidence has found that Chinese herbal medicine (CHM) can effectively treat chronic disease; nevertheless, it is unclear if adding CHM to the routine management of T2D would modify sequent risk of hearing loss. This cohort-based case-control study was conducted to address this issue. First, a total of 64,418 subjects aged 20–70 years, diagnosed with T2D between 2002 and 2011, were extracted from a nationwide health claims database. Among them, we identified 4516 cases of hearing loss after T2D by the end of 2013. They were then randomly matched to 9032 controls without hearing loss at a 1:2 ratio. Following conditional logistic regression, we found the addition of CHM to conventional care reduced the risk of developing hearing loss, with an adjusted odds ratio of 0.75 (95% confidence interval: 0.70–0.83). Specifically, taking CHM products for at least two years benefits T2D patients in lowering sequent risk of hearing loss. The findings herein implicated that integrating CHM into conventional care substantially correlated to lower risk of hearing loss for T2D patients, but further basic research is needed to secure the application of finished herbal products. Full article

In the midst of rigorous market rivalry, enhancing a company’s competitiveness and operational efficiency in an era of rapid IT advancement is a pressing concern for business leaders. The National Big Data Comprehensive Zone (BDCZ) pilot scheme, instituted by the Chinese government, systematically addresses seven core objectives, encompassing data resource management, sharing and disclosure, data center consolidation, application of data resources, and the circulation of data elements. This policy initiative aims to bolster the establishment of information infrastructure through big data applications, facilitate the influx and movement of talent, and propel corporate sustainable growth. Utilizing a quasi-natural experiment approach, we assess the pilot policy’s influence on the digital-intelligent transformation (DIT) of manufacturing companies from a green innovation ecosystem perspective, employing datasets from 2010 to 2022, and methodologies such as Difference-in-Differences (DID), Synthetic Differences-in-Differences (SDID), and Propensity Score Matching-DID (PSM-DID). The findings indicate that the BDCZ initiative significantly fosters DIT in manufacturing companies. The policy’s establishment confers benefits, including access to increased government support and innovation capital, thereby enhancing the sustainability of green innovation efforts. It also strengthens corporate collaboration, engendering synergistic benefits that improve regional economic progression and establish a conducive environment for digital development, ultimately enhancing the regional innovation ecosystem. The pilot policy’s impact varies across entities, with more profound effects observed in developed financial markets compared to underdeveloped ones. Additionally, non-state-owned companies exhibit a greater response to BDCZ policy interventions than their state-owned counterparts. Moreover, manufacturing bussiness with a higher proportion of executive shareholding are more substantially influenced by the BDCZ. This article fills the research gap by using the quasi-natural experiment of BDCZ to test the impact on DIT of companies and provides inspiration for local governments to mobilize the enthusiasm of manufacturing companies for DIT. Full article

In this paper, an ultra-thin logo antenna (LGA) operating in multiple frequency bands for Internet of Vehicles (IoVs) applications was proposed. The designed antenna can cover five frequency bands, 0.86–1.01 GHz (16.0%) for LoRa communication, 1.3–1.36 GHz (4.6%) for GPS, 2.32–2.71 GHz (16.3%) for Bluetooth communication, 3.63–3.89 GHz (6.9%) for 5G communication, and 5.27–5.66 GHz (7.1%) for WLAN, as the simulation indicated. The initial antenna started with a modified coplanar waveguide (CPW)-fed circular disk monopole radiator. To create extra current paths and further excite other modes, the disk was hollowed out into the shape of the car logo of the Chinese smart EV brand XPENG composing four rhombic parasitic patches. Next, four triangular parasitic patches were inserted to improve the impedance matching of the band at 5.6 GHz. Finally, four metallic vias were loaded for adjusting resonant points and the return loss reduction. Designed on a flexible substrate, the antenna can easily bend to a certain degree in complex vehicular communication for IoV. The measured results under horizontal and vertical bending showed the LGA can operate in a bending state while maintaining good performance. The proposed LGA addresses the issue of applying one single multi-band antenna to allow vehicles to communicate over several channels, which relieves the need for a sophisticated antenna network. Full article

Medical imaging description and disease diagnosis are vitally important yet time-consuming. Automated diagnosis report generation (DRG) from medical imaging description can reduce clinicians’ workload and improve their routine efficiency. To address this natural language generation task, fine-tuning a pre-trained large language model (LLM) is cost-effective and indispensable, and its success has been witnessed in many downstream applications. However, semantic inconsistency of sentence embeddings has been massively observed from undesirable repetitions or unnaturalness in text generation. To address the underlying issue of anisotropic distribution of token representation, in this study, a contrastive learning penalized cross-entropy (CLpCE) objective function is implemented to enhance the semantic consistency and accuracy of token representation by guiding the fine-tuning procedure towards a specific task. Furthermore, to improve the diversity of token generation in text summarization and to prevent sampling from unreliable tail of token distributions, a diversity contrastive search (DCS) decoding method is designed for restricting the report generation derived from a probable candidate set with maintained semantic coherence. Furthermore, a novel metric named the maximum of token repetition ratio (maxTRR) is proposed to estimate the token diversity and to help determine the candidate output. Based on the LLM of a generative pre-trained Transformer 2 (GPT-2) of Chinese version, the proposed CLpCE with DCS (CLpCEwDCS) decoding framework is validated on 30,000 desensitized text samples from the “Medical Imaging Diagnosis Report Generation” track of 2023 Global Artificial Intelligence Technology Innovation Competition. Using four kinds of metrics evaluated from n-gram word matching, semantic relevance, and content similarity as well as the maxTRR metric extensive experiments reveal that the proposed framework effectively maintains semantic coherence and accuracy (BLEU-1, 0.4937; BLEU-2, 0.4107; BLEU-3, 0.3461; BLEU-4, 0.2933; METEOR, 0.2612; ROUGE, 0.5182; CIDER, 1.4339) and improves text generation diversity and naturalness (maxTRR, 0.12). The phenomenon of dull or repetitive text generation is common when fine-tuning pre-trained LLMs for natural language processing applications. This study might shed some light on relieving this issue by developing comprehensive strategies to enhance semantic coherence, accuracy and diversity of sentence embeddings. Full article

Quantifier-Negation sentences such as all teachers did not use Sandy’s car are known to allow an inverse scope interpretation in English. However, there is a lack of experimental evidence to determine whether this interpretation is allowed in equivalent sentences in Japanese and Chinese. To address this issue, this study conducted a sentence–picture matching truth value judgment experiment in both Japanese and Chinese. The data suggested that Japanese Quantifier-Negation sentences do allow inverse scope readings, which suggests that the subject may be interpreted within the scope of negation. In contrast, Chinese Quantifier-Negation sentences prohibit inverse scope readings, which is in accordance with the strong scope rigidity consistently observed in this language. This paper also discussed how to develop a valid experiment for investigating scope ambiguities. Full article

In the field of Chinese-named entity recognition, recent research has sparked new interest by combining lexical features with character-based methods. Although this vocabulary enhancement method provides a new perspective, it faces two main challenges: firstly, using character-by-character matching can easily lead to conflicts during the vocabulary matching process. Although existing solutions attempt to alleviate this problem by obtaining semantic information about words, they still lack sufficient temporal sequential or global information acquisition; secondly, due to the limitations of dictionaries, there may be words in a sentence that do not match the dictionary. In this situation, existing vocabulary enhancement methods cannot effectively play a role. To address these issues, this paper proposes a method based on lexical information and spatial features. This method carefully considers the neighborhood and overlap relationships of characters in vocabulary and establishes global bidirectional semantic and temporal sequential information to effectively address the impact of conflicting vocabulary and character fusion on entity segmentation. Secondly, the attention score matrix extracted by the point-by-point convolutional network captures the local spatial relationship between characters without fused vocabulary information and characters with fused vocabulary information, aiming to compensate for information loss and strengthen spatial connections. The comparison results with the baseline model show that the SISF method proposed in this paper improves the F1 metric by 0.72%, 3.12%, 1.07%, and 0.37% on the Resume, Weibo, Ontonotes 4.0, and MSRA datasets, respectively. Full article

Spatial language understanding (SLU) is an important task in the field of information extraction, and it involves complex spatial analyses and reasoning processes. Unlike English SLU, in the case of the Chinese language, there may be some language-specific challenges, such as the phenomenon of polysemia and the substitution of synonyms. In this work, we explore Chinese SLU by taking advantage of large language models. Inspired by recent chain-of-thought (CoT) strategies, in this study, we propose the Spatial-CoT template to help improve LLMs’ reasoning abilities in order to deal with the challenges of Chinese SLU. Spatial-CoT offers LLMs three steps of instructions from different perspectives, namely, entity extraction, context analysis, and common knowledge analysis. We evaluate our framework on the Chinese SLU dataset SpaCE, which contains three subtasks: abnormal spatial semantics recognition, spatial role labeling, and spatial scene matching. The experimental results show that our Spatial-CoT outperforms vanilla prompt learning on ChatGPT and achieves competitive performance in comparison with traditional supervised models. Further analysis revealed that our method could address the phenomenon of polysemia and the substitution of synonyms in Chinese spatial language understanding. Full article

Few-shot named entity recognition requires sufficient prior knowledge to transfer valuable knowledge to the target domain with only a few labeled examples. Existing Chinese few-shot named entity recognition methods suffer from inadequate prior knowledge and limitations in feature representation. In this paper, we utilize enhanced Span and Label semantic representations for Chinese few-shot Named Entity Recognition (SLNER) to address the problem. Specifically, SLNER utilizes two encoders. One encoder is used to encode the text and its spans, and we employ the biaffine attention mechanism and self-attention to obtain enhanced span representations. This approach fully leverages the internal composition of entity mentions, leading to more accurate feature representations. The other encoder encodes the full label names to obtain label representations. Label names are broad representations of specific entity categories and share similar semantic meanings with entities. This similarity allows label names to offer valuable prior knowledge in few-shot scenarios. Finally, our model learns to match span representations with label representations. We conducted extensive experiments on three sampling benchmark Chinese datasets and a self-built food safety risk domain dataset. The experimental results show that our model outperforms the F1 scores of 0.20–6.57% of previous state-of-the-art methods in few-shot settings. Full article

Address parsing is a crucial task in natural language processing, particularly for Chinese addresses. The complex structure and semantic features of Chinese addresses present challenges due to their inherent ambiguity. Additionally, different task scenarios require varying levels of granularity in address components, further complicating the parsing process. To address these challenges and adapt to low-resource environments, we propose CapICL, a novel Chinese address parsing model based on the In-Context Learning (ICL) framework. CapICL leverages a sequence generator, regular expression matching, BERT semantic similarity computation, and Generative Pre-trained Transformer (GPT) modeling to enhance parsing accuracy by incorporating contextual information. We construct the sequence generator using a small annotated dataset, capturing distribution patterns and boundary features of address types to model address structure and semantics, which mitigates interference from unnecessary variations. We introduce the REB–KNN algorithm, which selects similar samples for ICL-based parsing using regular expression matching and BERT semantic similarity computation. The selected samples, raw text, and explanatory text are combined to form prompts and inputted into the GPT model for prediction and address parsing. Experimental results demonstrate significant achievements of CapICL in low-resource environments, reducing dependency on annotated data and computational resources. Our model’s effectiveness, adaptability, and broad application potential are validated, showcasing its positive impact in natural language processing and geographical information systems. Full article

The Gaofen-1 (GF-1) and Gaofen-6 (GF-6) satellites have acquired many GF-1 and GF-6 wide-field-view (WFV) images. These images have been made available for free use globally. The GF-1 WFV (GF-1) and GF-6 WFV (GF-6) images have rational polynomial coefficients (RPCs). In practical applications, RPC corrections of GF-1 and GF-6 images need to be completed using the rational function model (RFM). However, can the accuracy of the rational function model satisfy practical application requirements? To address this issue, a geometric accuracy method was proposed in this paper to evaluate the accuracy of the RFM of GF-1 and GF-6 images. First, RPC corrections were completed using the RFM and refined RFM, respectively. The RFM was constructed using the RPCs and Shuttle Radar Topography Mission (SRTM) 90 m DEM. The RFM was refined via affine transformation based on control points (CPs), which resulted in a refined RFM. Then, an automatic matching method was proposed to complete the automatic matching of GF-1/GF-6 images and reference images, which enabled us to obtain many uniformly distributed CPs. Finally, these CPs were used to evaluate the geometric accuracy of the RFM and refined RFM. The 14th-layer Google images of the corresponding area were used as reference images. In the experiments, the advantages and disadvantages of BRIEF, SIFT, and the proposed method were first compared. Then, the values of the root mean square error (RSME) of 10,561 Chinese, French, and Brazilian GF-1 and GF-6 images were calculated and statistically analyzed, and the local geometric distortions of the GF-1 and GF-6 images were evaluated; these were used to evaluate the accuracy of the RFM. Last, the accuracy of the refined RFM was evaluated using the eight GF-1 and GF-6 images. The experimental results indicate that the accuracy of the RFM for most GF-1 and GF-6 images cannot meet the actual use requirement of being better than 1.0 pixel, the accuracy of the refined RFM for GF-1 images cannot meet practical requirement of being better than 1.0 pixel, and the accuracy of the refined RFM for most GF-6 images meets the practical requirement of being better than 1.0 pixel. However, the RMSE values that meet the requirements are between 0.9 and 1.0, and the geometric accuracy can be further improved. Full article