This study systematically analyzed artificial intelligence (AI)-generated content text-to-image technology in product modeling innovation and built a systematic evaluation framework for enterprise design management. While AI quickly generates concepts, enterprises lack standard evaluation criteria and struggle to screen visual solutions suitable for engineering and manufacturing. By coding eight research documents, the study developed a double-loop collaborative mechanism, evaluating tools across six dimensions such as manufacturing adaptability and semantic matching. AI-generated content excels in early creative divergence but causes structural distortion and high management costs, though multi-modal systems show the best stability and deployment readiness. AI must evolve from a basic visual tool into a digital design infrastructure integrating engineering parameters and requiring standardized manual supervision to truly upgrade industrial design.
The authors conducted a bibliometric analysis of 765 papers on movie and television recommendation systems published between 2015 and 2025. The goal was to systematically chart the field's evolution from precise recommendations to personalized services while exposing key gaps in measuring diversity, fairness, and interpretability. They evaluated relevant literature from the Web of Science Core Collection and Scopus databases, using Bibliometrix and VOSviewer tools to map publication trends, global coauthorship clusters, and co-citation networks. The analysis revealed a 21.54% annual publication growth rate, indicating a distinct methodological shift from traditional collaborative filtering toward dynamic deep learning models and multimodal content fusion. These findings mean that modern systems are evolving from pure matching algorithms into human-centric service ecosystems, which requires urgent advances in real-time continuous feedback loops, cross-modal semantic understanding, and robust privacy protection.
This paper analyzes the impact of urban-level digital business environments (DBEs) on the performance of small and medium-sized enterprises (SMEs). Prior literature heavily equates digitalization with internal corporate transformations while ignoring the external digital institutional environment and its specific transmission mechanisms. This paper uses city-enterprise matching panel data of Chinese A-share SMEs from 2016 to 2023. It constructs a two-way fixed effect model and an intermediary effect model for empirical testing. The benchmarks reveal that an optimized DBE significantly boosts SME profitability, growth capability, and operational efficiency. Mechanistically, this occurs by easing financing constraints and intensifying corporate innovation investments. Heterogeneity tests show these benefits are most pronounced for private, high-tech, and central/western regional enterprises. Ultimately, this means DBEs act as vital institutional supports that transform urban governance modes, offering practical paths to foster sustainable, inclusive SME growth.
This study evaluated how AI intelligent financial systems affect cost control efficiency in small and medium-sized enterprises (SMEs). It was conducted because existing literature focuses heavily on macro-level adoption, ignoring micro-level workflows and leaving managers unsure about actual investment returns. A theoretical model based on decision delay, human-machine synergy, and investment boundaries was tested using survey data from 218 SMEs in the Yangtze River Delta. Results showed that efficiency depends on reducing delays, accurate cost prediction, and rational technical input, with prediction accuracy having the strongest impact. Data also proved an inverted U-shaped boundary for AI investment, where both under-investment and over-investment decrease net income. This means digital transformation value is jointly shaped by technology, organizational capability, and resource constraints, proving SMEs must identify optimal investment scales rather than just buying advanced systems.
This study constructed a closed-loop governance framework integrating learning behavior and academic performance big data with generative AI writing feedback. This was done because existing intelligent feedback tools focus on surface-level text generation and ignore learning behaviors, failing to explain differentiated outcomes across student ability levels. The research analyzed 1,248 sophomores in a 16-week online English writing course, utilizing behavior storage, diagnosis, scheduling, and calibration modules to dynamically adjust feedback granularity based on revision lag and absorption rates. It found an inverted U-shaped relationship between feedback length and score improvement, showing that tailoring intervention frequency and text length to specific student abilities significantly enhances writing performance. This means generative AI must evolve beyond a simple text correction tool into a standardized teaching component governed by risk thresholds to provide sustainable, personalized support.
