宜宾职业技术学院

Rapid and accurate evaluation of tea quality using artificial intelligence (AI) algorithms is crucial for effective quality management. Existing algorithms primarily focus on one-dimensional data (typically the appearance of dry tea) for classification or grading, often overlooking the complexity of tea flavor. Here, we developed Tea MvMtNet, a multi-view, multi-task deep neural network that processes visual data from dry tea appearance, infusion color, and waste tea to assess all five flavor factors, including taste and aroma, and to classify teas into distinct categories. The model incorporates an attention-enhanced bidirectional feature pyramid network for multi-scale fusion, extracting both global and local details from each view. The semantic enhancement module incorporates prior knowledge, leveraging cross-modal attention and graph attention networks to refine visual features for accurate prediction of flavor factors. Additionally, the multi-view fusion dynamically aggregates the three visual features to predict aroma, taste, and tea category. To optimize task performance and address data imbalance, we develop a multi-task balanced focal regression loss function for multi-task learning. Our approach achieved a classification accuracy of 98.11% and demonstrated superior precision in evaluating dry tea appearance (average absolute error percentage: 19.60%), infusion color (28.40%), aroma (26.48%), taste (31.67%), waste tea (17.20%), and overall score (17.23%), demonstrating that the accuracy of the model was comparable to that of senior tea evaluators. Furthermore, the overall score achieved an R2 value of 0.8359, with the best performance for aroma and taste (R2 > 0.75). These results highlight Tea MvMtNet as a comprehensive and precise tool for automated tea quality management.

This study developed an optimized processing strategy for γ-aminobutyric acid (GABA) instant white tea (GABA-IT) using GABA-enriched white tea as raw material, systematically characterizing its chemical composition and volatile profile. In contrast to the conventional instant tea production process, this work integrates response surface methodology with spray-drying parameter optimization. This integrated approach enables the simultaneous enhancement of functional components and sensory quality. A response surface design was employed to refine the extraction and spray-drying variables following preliminary single-factor trials, and the optimal parameter combination was subsequently determined (40% ethanol concentration, material-to-liquid ratio of 1:15, extraction time of 3 days, atomization speed of 300 rpm, and inlet temperature of 120 °C); the resulting GABA-IT exhibited significantly improved quality characteristics. Specifically, the GABA content increased by 209% (reaching 4.42 mg/g), and theanine, catechins, and caffeine were enriched by 200–300%. Regarding volatile profiles, processing led to a reduction in esters but an increase in aldehydes and hydrocarbons. Relative odor activity value (rOAV) analysis revealed that epoxy-β-ionone and linalool were the key contributors to the characteristic aroma of GABA-IT. Collectively, this study demonstrates the technical feasibility of producing GABA-rich instant tea with enhanced functional components and improved sensory quality, providing practical guidance for the large-scale industrial production of functional tea beverages.

This study synthesizes empirical evidence on AI-supported skill assessment systems in higher vocational education through a systematic review and meta-analysis. Despite growing interest in generative AI within higher education, empirical research on AI-enabled assessment remains fragmented and methodologically uneven, particularly in vocational contexts. Following PRISMA 2020 guidelines, 27 peer-reviewed empirical studies published between 2010 and 2024 were identified from major international and Chinese databases and included in the analysis. Using a random-effects model, the meta-analysis indicates a moderate positive association between AI-supported assessment systems and skill-related learning outcomes (Hedges’ g = 0.72), alongside substantial heterogeneity across study designs, outcome measures, and implementation contexts. Subgroup analyses suggest variation across regional and institutional settings, which should be interpreted cautiously given small sample sizes and diverse methodological approaches. Based on the synthesized evidence, the study proposes a conceptual AI-supported skill assessment framework that distinguishes empirically grounded components from forward-looking extensions related to generative AI. Rather than offering prescriptive solutions, the framework provides an evidence-informed baseline to support future research, system design, and responsible integration of generative AI in higher education assessment. Overall, the findings highlight both the potential and the current empirical limitations of AI-enabled assessment, underscoring the need for more robust, theory-informed, and transparent studies as generative AI applications continue to evolve.

Photocatalysis and piezocatalysis hold great promise for pollutant degradation, yet their practical applications are often limited by poor visible-light utilization and inefficient charge separation. Here, a BaTiO3@TiO2_ x (BT@TiO2_ x) core-shell heterojunction was fabricated through a single-step hydrothermal method using a Ti3+ precursor. X-ray diffraction (XRD) and transmission electron microscopy (TEM) confirmed the crystalline phases and core-shell architecture, while evidence for oxygen-vacancy-related defects was obtained from X-ray photoelectron spectroscopy (XPS) and electron paramagnetic resonance (EPR). Under simultaneous light irradiation and ultrasonic vibration, BT@TiO2_ x degraded 99.01% of Rhodamine B (RhB) within 80 min. The rate constant (k) was 5.39 & times; 10_2 min_ 1, indicating a 7.8-fold increase over pristine BaTiO3. Oxygen vacancies introduce defect-related states into the TiO2_ x shell lattice, thereby enhancing visible-light absorption and facilitating charge separation. The piezoelectric field of the BaTiO3 core, together with the interfacial electronic configuration at the core-shell interface, promotes charge separation and directional carrier transport. These results suggest that the enhanced catalytic activity can be attributed to the synergistic effects of oxygen-vacancy engineering, interfacial heterojunction formation, and the piezoelectric field of the BaTiO3 core. This work presents an effective approach for designing efficient piezo-photocatalysts for environmental remediation.

Traditional fermented sausages are widely consumed in China but often face quality and safety risks due to contamination by spoilage and pathogenic microorganisms during natural fermentation. This study investigated the effect of Litsea cubeba essential oil (LCEO) on the microbial community structure and volatile flavor compounds (VFCs) of fermented sausages using multi-omics approaches, including 16S/ITS amplicon sequencing and HS-SPME-GC-MS. The results showed that LCEO significantly altered the microbial community, increasing the relative abundance of Lactobacillus from 4.65% to 76.22% and Debaryomyces from 11.95% to 22.35%, while reducing Staphylococcus and Aspergillus by 47.01% to 4.55% and 37.28% to 17.09%, respectively. A total of 695 VFCs were identified, with terpenoids and hydrocarbons being the most abundant. A differential analysis revealed that LCEO enhanced the relative content of key aroma-active compounds, including ethyl esters and aldehydes, with odor activity values exceeding 1 for fruity-related compounds such as decanoic acid, ethyl ester and nonanoic acid, methyl ester. Strikingly, the elevated levels of these fruity esters translated into significantly higher sensory scores for odor (p < 0.05) in LCEO-treated sausages, as panelists consistently perceived a pronounced fruity and fresh aroma profile, ultimately leading to a superior overall acceptability compared to the control. A correlation analysis indicated strong positive associations between Lactobacillus, Debaryomyces, and major flavor-contributing esters. In conclusion, LCEO potentially reduces the microbial risks and improves the flavor profile of fermented sausages, making it a promising natural additive for meat fermentation.