RESUMO
Social-network-based recommendation algorithms leverage rich social network information to alleviate the problem of data sparsity and boost the recommendation performance. However, traditional social-network-based recommendation algorithms ignore high-order collaborative signals or only consider the first-order collaborative signal when learning users' and items' latent representations, resulting in suboptimal recommendation performance. In this paper, we propose a graph neural network (GNN)-based social recommendation model that utilizes the GNN framework to capture high-order collaborative signals in the process of learning the latent representations of users and items. Specifically, we formulate the representations of entities, i.e., users and items, by stacking multiple embedding propagation layers to recursively aggregate multi-hop neighborhood information on both the user-item interaction graph and the social network graph. Hence, the collaborative signals hidden in both the user-item interaction graph and the social network graph are explicitly injected into the final representations of entities. Moreover, we ease the training process of the proposed GNN-based social recommendation model and alleviate overfitting by adopting a lightweight GNN framework that only retains the neighborhood aggregation component and abandons the feature transformation and nonlinear activation components. The experimental results on two real-world datasets show that our proposed GNN-based social recommendation method outperforms the state-of-the-art recommendation algorithms.
Assuntos
Algoritmos , Redes Neurais de Computação , Aprendizagem , Registros , Rede SocialRESUMO
This paper provides a scheme for realizing high-accuracy, low-cost resonant fiber optical gyroscopes (RFOGs) with single-mode (SM) fibers. The temperature-related polarization noise, which deteriorates the bias stability of the RFOG, has been suppressed thanks to a much longer birefringence beat length of SM fibers. It is also helpful to improve the theoretical sensitivity of the RFOG because of the very low loss of the SM fiber components. A transmissive 39-m-long SM fiber ring resonator (FRR) is fabricated. The measured finesse is 58.7, and its total loss is calculated to be less than 0.3 dB. Using this SM-FRR as the sensing element, an RFOG is then set up that achieves an angular random walk of 0.0039∘/âh and a bias stability of 0.05°/h. The experimental results show that the SM-FRR scheme would be a good candidate for producing high-accuracy and low-cost RFOGs.
RESUMO
OBJECTIVE: To examine the oral health status among the orphan children and adolescents of Chongqing and assist in planning of the oral health programs. METHODS: According to the third national oral health investigation of epidemiology, the dental caries, gingival bleeding and calculus was examined among 317 orphan living in Chongqing, by cluster sampling. The statistical software SPSS 17.0 was used for the data analysis. RESULTS: In primary and permanent teeth, the prevalence of dental caries and mean DMFT (dmft) were found to be 50.00%, 1.94 +/- 2.81 and 39.53%, 0.90 +/- 1.38. There were no significant difference between female and male (P>0.05). But there was significant difference of the prevalence of permanent tooth caries between the age group under 12 (include 12) and above 12 (P < 0.05). It was also found that 35.25% of the orphan children and adolescents had caries in the first permanent molar, and there was significant difference between female and male (P < 0.05). Neither of the caries teeth above was treated nor the healthy teeth were sealed. The prevalence rate of gingival bleeding was 78.22%, and the calculus rate was 67.66%. CONCLUSION: The oral status of the orphan is poor, and this community has experienced a low utilization of preventive or therapeutic oral health services. In the future work, orphans should be considered as a priority group when make plans for oral health care.
