With the rapid iteration of artificial intelligence technology, model training has shifted from the stage of general-scale expansion to a new stage driven by industrial applications and supported by high-quality data. The type, quality and supply capacity of data directly determine the industry adaptability and practical effectiveness of models. Data demands vary across the full life cycle of large models, and there are still shortcomings in the supply of high-quality datasets for industrial scenarios. Based on this, the focus is placed on the full stages of large model pre-training, supervised fine-tuning, reinforcement alignment, and engineering application. This paper sorts out the data demands of each stage and the evolutionary trend from “scale-oriented” to “quality-oriented”. The evolving trends of data needs at each stage are analyzed, the disparities in high-quality data supply between international and domestic standards are compared, and the shortcomings of China’s utilization of public data, data openness, and annotation ecosystems are examined. Consequently, targeted measures are proposed to optimize data supply, aiming to break through the bottleneck of training data for large models and support the advancement of the artificial intelligence industry.

High-quality datasets have become a critical foundation for supporting the development of artificial intelligence and the intelligent transformation of industries. Based on this, the study first analyzes the concept and current development status of high-quality datasets. Secondly, it examines the construction requirements and core issues of different types of datasets. Thirdly, it investigates key stages and pathways in the process of building enterprise-level high-quality datasets, analyzing the construction approaches for high-quality datasets in the healthcare, energy, and industrial sectors. Finally, it explores future development trends of high-quality datasets.

Currently, the recognition of data elements as assets has become a central concern in the industry. Drawing on practical observations of data governance in the securities industry, this study systematically reviews the policy evolution and industry landscape of high-quality dataset construction, and provides an in-depth analysis of the structural bottlenecks and institutional challenges encountered by the securities industry in building investment datasets and recognizing data resources in financial statements. By integrating perspectives from data governance, asset recognition, and accounting measurement, this paper constructs a theoretical framework covering organizational optimization, data standardization, compliance and rights-confirmation mechanisms, and cost-benefit measurement methodologies. It further proposes a phased implementation roadmap for data resource accounting recognition, aiming to offer theoretical insights and decision-making references for financial institutions’ value-oriented transformation of data elements.

Building high-quality datasets for AI applications often faces four practical challenges: unclear alignment with business goals, fragmented implementation, limited technical infrastructure, and excessive annotation costs, this paper presents a methodology that addresses these issues through a three-layer framework—demand mapping, intelligent governance, and value realization—implemented on China Telecom’s Knowledge Service Platform. The methodology has been validated in high-end equipment manufacturing and consumer goods industries, cutting dataset construction time, offering a practical pathway for enterprise data asset development in the era of data marketization.

This paper focuses on the demand for high-quality datasets in large model development and proposes an enterprise-level platform methodology covering resource management, automated processing, quality evaluation, version traceability, and secure sharing. By integrating visual workflows, high-performance filtering, hybrid scheduling, and a multidimensional evaluation framework, the platform addresses key challenges in large-scale data processing, including efficiency, compliance, and governance, and provides a practical solution for high-quality dataset construction.

With the rapid evolution of generative artificial intelligence technology, data quality has become the core bottleneck restricting the performance of industry-scale large language models. Telecom operators possess ZB-scale cross-domain data, providing inherent resource advantages for training vertical large language models. However, raw communications data generally faces issues such as multi-source heterogeneity, high redundancy, and scarcity of long-tail samples, which limits its effectiveness when directly applied to model training. To address this, the system proposes a high-quality dataset construction method encompassing “collection—governance-annotation-evaluation,” featuring key technologies such as deep semantic compression, long-tail data synthesis based on Wasserstein Generative Adversarial Network (WGAN) and Long Short-Term Memory (LSTM) network, domain ontology construction, and human-machine collaborative annotation. Meanwhile, a general and specialized knowledge data coordination mechanism is designed to effectively mitigate catastrophic forgetting during industry fine-tuning. Practice has proved that this method is effective and can provide reference for the construction of high-quality datasets in the telecommunications industry.

To address the challenge of lacking effective metrological evaluation methods for the output values of deep learning object detection algorithms due to their lack of interpretability, this study constructs a technical framework for algorithm traceability from a metrological perspective. Using ship detection in Synthetic Aperture Radar (SAR) images as a typical application scenario, the definition and uncertainty sources of algorithm traceability are clarified, and a traceability technical path based on standard datasets together with a continuous comparison chain is proposed. To meet the metrological requirements of algorithm traceability, a quality evaluation indicator system and standardized testing methods for standard datasets are established. This study has achieved quantitative evaluation and traceability of the performance metrics generated by SAR image-based ship detection algorithms. It provides a benchmark for the reliable evaluation of artificial intelligence algorithms and is of great significance for advancing the development of a standardized evaluation system.

As cutting-edge achievements in the field of artificial intelligence, large language models have drawn significant attention from both academia and industry regarding their associated ethical issues. Consequently, the number of ethical evaluation datasets for large language models in the Chinese context has gradually increased, presenting substantial value for in-depth research. However, the current lack of systematic review and analysis of such datasets makes it difficult for researchers to accurately select suitable datasets and effectively identify shortcomings in existing resources. This paper examines 50 Chinese ethical evaluation datasets for large language models released between August 2021 and March 2025. It conducts a comprehensive comparative analysis covering release dates, creation details, content information, open-source situation, domains covered, and ethical scenarios. This study aims to provide direction for optimizing and constructing future datasets.

This paper focuses on three core dimensions in developing high-quality datasets for embodied AI: data collection and source governance, quality evaluation and assessment mechanisms, and sharing, circulation and application support. It systematically reviews the current status of policies, regulations and standards in major economies worldwide. The European Union has established a compliance baseline through high-level legislation. The United States of America has shaped its technological ecosystem led by leading enterprises. China has promoted systematic development via top-level strategic guidance and standardized layout. Going forward, China should accelerate high-level legislation, improve the quality evaluation standards for embodied AI datasets, strengthen data infrastructure, and refine the data circulation system to support the high-quality development of the embodied AI industry.

Artificial Intelligence (AI) acts as a significant driving force for new-type industrialization, leading a new round of technological revolution and industrial transformation. The continuous development of AI technologies such as deep learning, reinforcement learning, and large models, along with the increasing emergence of application scenarios like visual quality inspection, production scheduling, and robot control, is driving the manufacturing industry towards a new stage of comprehensive intelligence. This paper focuses on the empowerment of industrial site automation through AI, reviews the integration of industrial optimization control methods, intelligent optimization control algorithms, AI technologies, and industrial optimization control, and introduces application explorations in some industries.

Traditional IP address-based identification technologies face challenges such as protocol stack layer misalignment and weak semantic capability, making it difficult to meet the service scheduling needs of application-level computing power services. To address these challenges, this paper proposes a core technical path for service semantic routing. The core idea involves encoding service requirements into routable semantic identifiers, enabling accurate identification and intelligent routing of computing power services directly at the application layer. This study analyzes the evolution trend of service semantic routing technology, constructs an identification foundation for resource-service collaboration, and elaborates on two service semantic routing schemes: a network-layer scheme based on Virtual IP Anycast and an application-layer scheme based on URL-PATH. The comparative analysis shows that the application-layer scheme based on URL-PATH has significant advantages in terms of flexibility, intuitiveness, and ease of deployment, making it better suited for the development needs of application-level computing power services.

In China’s domestic violence governance practices, there exist several practical issues, such as limited avenues for relief, difficulty in gathering evidence, challenge in submitting evidence, complexities in substantiating acts of domestic violence, and insufficient safeguards for personal safety protection orders. The development of smart wearable devices offers new perspectives and opportunities for empowering domestic violence governance. From the perspective of value foundation, respecting and safeguarding human rights, promoting substantive justice through active justice, and promoting technology for goodness and a better life are important value bases for empowering domestic violence governance through smart wearable devices. In terms of functional application scenarios, smart wearable devices can play an active role in evidence collection and preservation, data monitoring and reasoning, danger warnings, and safety assurance. Therefore, legislative efforts should be improved in areas such as the evidentiary validity of electronic data and technical standards for evidence gathering. Additionally, the therapeutic function of family justice in healing family relationships should be fully utilized, with various methods employed to alleviate and resolve the psychological distress caused by domestic violence. At the same time, special attention must be paid to protecting the personal information and privacy of individuals involved during the process of empowering domestic violence governance through smart wearable devices. Where necessary, rights such as the “right to deletion” and the “right to be forgotten” should be introduced. In this way, smart wearable devices can effectively empower domestic violence governance within the framework of the rule of law.