Knowledge tracing models based on deep learning can achieve impressive predictive performance by leveraging attention mechanisms. However, there still exist two challenges in attentive knowledge tracing (AKT): First, the mechanism of classical models of AKT demonstrates relatively low attention when processing exercise sequences with shifting knowledge concepts (KC), making it difficult to capture the comprehensive state of knowledge across sequences. Second, classical models do not consider stochastic behaviors, which negatively affects models of AKT in terms of capturing anomalous knowledge states. This article proposes a model of AKT, called Enhancing Locality for Attentive Knowledge Tracing (ELAKT), that is a variant of the deep KT model. The proposed model leverages the encoder module of the transformer to aggregate knowledge embedding generated by both exercises and responses over all timesteps. In addition, it uses causal convolutions to aggregate and smooth the states of local knowledge. The ELAKT model uses the states of comprehensive KCs to introduce a prediction correction module to forecast the future responses of students to deal with noise caused by stochastic behaviors. The results of experiments demonstrated that the ELAKT model consistently outperforms state-of-the-art baseline KT models.

With the expansion of microservice-based applications over time, the number of microservices rises, resulting in an augmentation of the volume of performance metrics. Consequently, selecting the appropriate performance metrics for anomaly detection becomes a critical challenge. Since these performance metrics are typically strongly correlated with timestamps, they form time series data comprising timestamp–value pairs. To address this, we propose SRdetector, a feature-enhanced Transformer-based model that adopts a time series forecasting approach to detect anomalies in microservices. Furthermore, we integrate a dynamic weight adjustment mechanism into the original Transformer attention mechanism to assign weights to different performance and temporal features. This enables the model to dynamically learn the significance of various features at different time intervals, effectively serving as a feature selection method for microservice performance metrics. Finally, anomaly detection in microservices is conducted by evaluating the predicted performance metric data based on confidence intervals.

In the digital age, social media significantly shapes public attitudes and emotional responses towards health interventions, such as HPV vaccination, which is critical in developing countries. This study employed a deep learning model to identify fine-grained emotions of 38,615 HPV-related tweets from 2016 to 2024, revealing significant shifts in public emotions. Notably, skepticism about vaccine commercialization motives heightened anger, while university outreach initiatives fostered positive emotions. Structural entropy analysis highlighted polarized emotional communication networks: the network of joy exhibited lower entropy with centralized information flow, whereas other emotions displayed higher entropy, fragmented dissemination, and enhanced cross-community communication efficiency. New communicators, such as campus accounts and music bloggers, played pivotal roles in spreading positive emotions, while individual bloggers in specific fields amplified negative emotions like anger, particularly in closed networks. This research underscores the intricate dynamics of online health communication and the need for targeted interventions to address stigma and enhance public awareness of HPV vaccination, providing valuable insights for future public health policy.