Abstract—Online learning platforms provide diverse course resources, but this often result in the issue of information overload. Learners always want to learn courses that are appropriate for their knowledge level and preferences quickly and accurately. Effective course recommendation plays a key role in helping learners select appropriate courses and improving the efficiency of online learning. However, when a user is enrolled in multiple courses, Existing course recommendation methods face the challenge in accurately recommending the target course that is most relevant to the user, because of the noise courses. In this paper, we propose a novel reinforcement learning model named Actor-Critic Hierarchical Reinforcement Learning (ACHRL). The model incorporates the Actor-Critic method to construct the profile reviser. This can remove noise courses and make personalized course recommendation effectively. Furthermore, we propose a policy gradient based on temporal difference error to reduce the variance in the training process, to speed up the convergence of the model, and improves the accuracy of the recommendation. We evaluate the proposed model on two real datasets, and the experimental results show that the proposed model is significantly outperforms the existing recommendation models (improving 3.77% to 13.66% in terms of HR@5).

The application of naturally ventilated pig buildings (NVPBs) with outdoor exercise yards is on the rise mainly due to animal welfare considerations, while the issue of emissions from the buildings to the surrounding environment is important. Since air pollutants are mainly transported by airflow, the knowledge on the airflow characteristics downwind the building is required. The objective of this research was to investigate airflow properties downwind of a NVPB with a roofed outdoor exercise yard for roof slopes of 5°, 15°, and 25°. Air velocities downwind a 1:50 scaled NVPB model were measured using a Laser Doppler Anemometer in a large boundary layer wind tunnel. A region with reduced mean air velocities was found along the downwind side of the building with a distance up to 0.5 m (i.e. 3.8 times building height), in which the emission concentration might be high. It was found that a smaller roof slope (i.e. 5° slope) resulted in a higher and shorter wake zone and thus a shorter air pollutant dispersion distance. It was concluded that a smaller roof slope could contribute to the dilution of air pollutants and a lower air pollutant concentration near the ground.

In order to reduce the waste of Akebia trifoliata peel and maximize its utilization. In this study, on the basis of single factor experiment and response surface method, the optimum technological conditions for the extraction of soluble dietary fiber from Akebia trifoliata peel by compound en-zyme method were obtained. The chemical composition, physical and chemical properties, structural characterization and biological activity of the purified soluble dietary fiber (AP-SDF) from the Akebia trifoliata peel were analyzed. We discovered that that the optimum yield was 20.87 % under the conditions of cellulase addition 600 U/g, enzymolysis time 100 min, solid-liquid ratio 1:24 g/mL and enzymolysis temperature 51 °C. At the same time, AP-SDF was a po-rous network structure cellulose type I acidic polysaccharose mainly composed of arabinoxylan (36.03%), galacturonic acid (27.40%) and glucose (19.00%), which possessed the structural charac-teristic peaks of the infrared spectra of polysaccharides, and the weight average molecular weight (Mw) was 95.52 kDa with good uniformity. In addition, AP-SDF exhibited high oil hold-ing capacity (15.11 g/g), good water holding capacity, swelling capacity, certain antioxidant ca-pacity in vitro, hypoglycemic activity in vitro of α-glucosidase inhibition and hypolipidemic ac-tivity in vitro of the binding ability of bile acids and cholesterol. These results will provide a theoretical basis for the development of functional products with antioxidant, hypoglycemic and hypolipidemic effects, it has certain application value in related industries.

Lodging is a common problem in maize production that seriously impacts yield, quality, and the capacity for mechanical harvesting. Evaluation of site-specific lodging risks requires establishment of a method for multi-year monitoring. In this study, spectral images collected by the Sentinel-2 satellite were processed to obtain three types of data: gray-level co-occurrence matrix texture (GLCM), vegetation indices (VIs), and spectral reflectance (SR). Lodging classification models were then established with Random Forest (RF) using each of the three data types separately (the GLCM, VI, and SR models) and in combination (SR+VI model, SR+GLCM model, VI+GLCM mod-el, and SR+VI+GLCM model). By gradually removing features with low importance scores from the SR+VI+GLCM model and analyzing the changes in the overall accuracy (OA), the optimal set of predictive variables was identified and used to construct the optimal model. A model built us-ing data from a single timepoint in 2021 was tested on data collected at a similar timepoint in 2019 and vice versa to assess interannual model generalizability. The results of this study demon-strate that for monitoring maize lodging, models constructed with a single feature type, the GLCM model had significantly lower accuracy compared to the VI and SR models. During certain growth stages, the model constructed with combined features had significantly higher accuracy in monitoring maize lodging compared to models constructed with a single feature. During the pro-cess of selecting the optimal predictive variables, it was found that the accuracy of the model did not increase as the number of predictive variables increased. The results show that the positive and negative validation models had an accuracy of 96.55% and 95.18%, with kappa values of 0.93 and 0.83, respectively. This indicates that the model has strong generality for the same repro-ductive stage between years. This study provides a detailed method for large-scale maize lodging monitoring, allowing for identification of optimal planting practices to reduce the probability of lodging and ultimately improving regional maize yield and quality.