The task of charge prediction is to predict the charge based on the fact description. Existing methods have a good effect on the prediction of high-frequency charges, but the prediction of low-frequency charges is still a challenge. Moreover, there exist some confusing charges that have relatively similar fact descriptions, which can be easily misjudged. Therefore, we propose a model with data augmentation and feature augmentation for few-shot charge prediction. Specifically, the model takes the text description as the input and uses the Mixup method to generate virtual samples for data augmentation. Then, the charge information heterogeneous graph is introduced, and a novel graph convolutional network is designed to extract distinguishability features for feature augmentation. A feature fusion network is used to effectively integrate the charge graph knowledge into the fact to learn semantic-enhanced fact representation. Finally, the semantic-enhanced fact representation is used to predict the charge. In addition, based on the distribution of each charge, a category prior loss function is designed to increase the contribution of low-frequency charges to the model optimization. The experimental results on real-work datasets prove the effectiveness and robustness of the proposed model.
In order for industrial and mining enterprises to fast detect the quality of fine mineral powders with far infrared emission, a simple testing model was set up according to the relationship between the emission intensity of powders and their surface temperature. The corresponding testing device was designed and assembled into three parts containing Constant Temperature Heating Part, Temperature Measuring Part and Sample Loading Part. By using the tourmaline mineral powders with far infrared emission as the research object and combining Fourier transform infrared spectroscopy, the calibration for the testing device was carried out. The results showed that the far infrared emission intensity of the tourmaline powders with different mining area and particle size could be judged. The testing results exhibited correct values when compared with those from FTIR measurements.
Most existing Web service recommendation models based on machine learning do not fully consider the high-order features interaction between users and services and with poor interpretability. In this paper, an Interpretable Web Service Recommendation model based on Disentangled Representation Learning (WSR-DRL) is proposed. First of all, to make full use of the service description information to improve the accuracy of Web service recommendation, the features representation of service name is obtained by using BERT model, and the local and global features representation of service description information is further obtained by combining 2-D CNN and Bi-LSTM. Then the disentangled convolution neural network is used to generate the high-order interaction features between users and services, and the neighborhood routing algorithm is used to mine the latent factors in these features. That improves the accuracy of Web service recommendation and make it interpretable. Finally, in order to verify the effectiveness of the model, several groups of experiments are carried out on real data sets. The experimental results show that compared with latest models such as DMF, DeepFM, DKN, GCMC, NDCG model and WSR-MGAT model, the WSR-DRL model proposed in this paper shows better performance on Precision@10, Recall@10, F1@10 and NDCG@10 evaluation metrics.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.