Jingdezhen Ceramic Institute

This paper proposes an innovative automated essay scoring(AES) and feedback generation method based on the LLaMA-3 model and Multi-task LoRA (M-LoRA) fine-tune technology, aimed at improving the accuracy of IELTS essay scoring and the quality of personalized feedback generation. Our approach consists of three key stages: multi-task supervised fine-tuning of LLaMA-3 model, designing of the reward model, and reinforcement learning model training based on fine-grained human feedback. Before the experiment, we collected a private dataset of 5,088 IELTS essays with expert-annotated feedback and used this dataset to train and fine-tune the entire model. Firstly, through multi-task supervised fine-tuning, we successfully captured features efficiently across the four key dimensions of IELTS essay scoring: Task Response, Coherence and Cohesion, Lexical Resource, and Grammatical Range and Accuracy, effectively addressing the issue of catastrophic forgetting in scoring tasks. Secondly, we designed and trained a reward model to optimize the ability to generate feedback by scoring the quality of the feedback. Finally, we further fine-tuned the generated feedback using a reinforcement learning strategy model based on fine-grained human feedback, making the feedback more refined and personalized. Our findings demonstrate significant improvements in both essay scoring and feedback generation, showcasing practical applications in real-world educational settings. This research highlights the limitations of current large language models in grasping the complexities of essay scoring, emphasizing the need for more effective methods like ours to advance this field.

This study investigates how gradient layer thickness affects the thermal response of CuSn10/316L stainless steel composites fabricated by selective laser melting (SLM). Three gradient architectures with varying layer thicknesses were designed and analyzed through both simulation and experiment. Simulations predicted that increasing interfacial layers would enhance thermal barrier performance. However, experimental results showed the opposite trend: samples with thinner layers and lower porosity (10.03%) exhibited poorer thermal barrier properties, while those with thicker layers and higher porosity (15.42%) performed best. This discrepancy reveals that porosity—governed by layer thickness—is the dominant factor controlling thermal behavior, outweighing the effect of interfacial density. Thicker layers promote heat accumulation and gas entrapment, increasing porosity and thermal resistance. The findings establish a clear pathway from gradient layer thickness to porosity evolution to thermal response, providing practical guidance for designing functionally graded materials with tailored thermal barrier performance.

The rapid development of archaeology in China has led to the excavation of numerous fragmented porcelain artifacts, for which adhesive materials play a critical role in conservation and restoration. The long-term stability of these adhesives directly affects the structural safety and visual integrity of restored objects. In this study, four adhesives widely used in Chinese conservation practice-epoxy resin Hezhong AAA, epoxy resin Hongxing 509, acrylic resin Paraloid B-72, and cyanoacrylate adhesive 502-were systematically investigated through simulated cyclic aging experiments. A multi-analytical approach was employed, including ultra-depth-of-field microscopy, CIE Lab* colorimetric analysis, pencil hardness testing, and Fourier transform infrared spectroscopy (FTIR). The results reveal distinct aging behaviors among different adhesive types. Epoxy resin adhesives exhibit high initial hardness and pronounced hardening during aging, with coating hardness increasing from the B range to the H range after 15 aging cycles; however, they also show significant yellowing, with total color differences (ΔE) exceeding 10 and dominated by increases in the b* parameter. Paraloid B-72 maintains excellent color stability throughout aging, with ΔE values consistently below 2, although it shows limited thermal stability and delayed physical stabilization. The cyanoacrylate adhesive 502 demonstrates rapid curing and minimal discoloration but undergoes embrittlement and interfacial debonding during aging, indicating reduced long-term bonding reliability. By correlating macroscopic performance evolution with molecular-level chemical changes, this study elucidates the aging mechanisms of commonly used restoration adhesives and provides a scientific basis for adhesive selection, risk assessment, and long-term preservation strategies in porcelain conservation.

Abstract This study investigates the factors that influence college students’ intention to use artificial intelligence generated content (AIGC) technology in design learning. An extended technology use and diffusion model is proposed and validated by integrating the artificial intelligence device use acceptance (AIDUA) model with the innovation diffusion theory (IDT). The present study collected data from 385 Chinese college students majoring in design through online surveys. The proposed model, which includes technology concerns, emotional acceptance, and behavioral transformation, was empirically tested using structural equation modeling (SEM) on data collected from students across different academic levels. The research findings suggest that, in the first stage, relative advantage and compatibility exert a significant and positive influence on both performance expectancy and effort expectancy. However, complexity negatively affects both. In the second stage (emotional acceptance), effort expectancy has a highly positive and significant influence on the adoption and diffusion. Conversely, the impact of performance expectancy on use and diffusion is not substantial. In the final stage (behavioral transformation), both social influence and individual innovation positively and significantly impact the use and diffusion of AIGC. Thus, the empirical results support the integration of AIDUA and IDT. This study provides a conceptual AIGC use and diffusion framework that other researchers can use to investigate AIGC-related topics in design learning.