Visual Question Answering (VQA) is a multimodal task that uses natural language to ask and answer questions based on image content. For multimodal tasks, obtaining accurate modality feature information is crucial. The existing researches on the visual question answering model mainly start from the perspective of attention mechanism and multimodal fusion, which will tend to ignore the impact of modal interaction learning and the introduction of noise information in the process of modal fusion on the overall performance of the model. This paper proposes a novel and efficient multimodal adaptive gated mechanism model, MAGM. The model adds an adaptive gate mechanism to the intra- and inter-modality learning and the modal fusion process. This model can effectively filter irrelevant noise information, obtain fine-grained modal features, and improve the ability of the model to adaptively control the contribution of the two modal features to the predicted answer. In intra- and inter-modality learning modules, the self-attention gated and self-guided-attention gated units are designed to filter text and image features’ noise information effectively. In modal fusion module, the adaptive gated modal feature fusion structure is designed to obtain fine-grained modal features and improve the accuracy of the model in answering questions. Quantitative and qualitative experiments on the two VQA task benchmark datasets, VQA 2.0 and GQA, proved that the method in this paper is superior to the existing methods. The MAGM model has an overall accuracy of 71.30% on the VQA 2.0 dataset and an overall accuracy of 57.57% on the GQA dataset.

Chinese long text classification plays a vital role in Natural Language Processing. Compared to Chinese short texts, Chinese long texts contain more complex semantic feature information. Furthermore, the distribution of these semantic features is uneven due to the varying lengths of the texts. Current research on Chinese long text classification models primarily focuses on enhancing text semantic features and representing Chinese long texts as graph-structured data. Nonetheless, these methods are still susceptible to noise information and tend to overlook the deep semantic information in long texts. To address the above challenges, this study proposes a novel and effective method called MACFM, which introduces a deep feature information mining method and an adaptive modal feature information fusion strategy to learn the semantic features of Chinese long texts thoroughly. First, we present the DCAM module to capture complex semantic features in Chinese long texts, allowing the model to learn detailed high-level representation features. Then, we explore the relationships between word vectors and text graphs, enabling the model to capture abundant semantic information and text positional information from the graph. Finally, we develop the AMFM module to effectively combine different modal feature representations and eliminate the unrelated noise information. The experimental results on five Chinese long text datasets show that our method significantly improves the accuracy of Chinese long text classification tasks. Furthermore, the generalization experiments on five English datasets and the visualized results demonstrate the effectiveness and interpretability of the MACFM model.

Multi-label text classification (MLTC) aims to assign the most appropriate label or labels to each input text. Previous studies have focused on mining textual information, ignoring the interdependence of labels and texts, thus leading to the loss of information about labels. In addition, previous studies have tended to focus on the single granularity of information in documents, ignoring the degree of inclination towards labels in different sentences in multi-labeled texts. In order to solve the above problems, this paper proposes a Label-Sentence Bi-Attention Fusion Network (LSBAFN) with multi-level feature extraction for mining multi-granularity information and label information in documents. Specifically, document-level and sentence-level word embeddings are first obtained. Then, the textual relevance of the labels to these two levels is utilized to construct sentence-level textual representations. Next, a multi-level feature extraction mechanism is utilized to acquire a sentence-level textual representation that incorporates contextual information and a document-level textual representation that reflects label features. Subsequently, the label-sentence bi-attention fusion mechanism is used to learn the feature relationships in the two text representations and fuse them. Label attention identifies text features related to labels from the document-level text representation, while sentence attention focuses on the tendency of sentences towards labels. Finally, the effective portion of the fused features is extracted for classification by a multi-layer perceptron. The experimental findings indicate that the LSBAFN can improve the effectiveness of the MLTC task. Compared with the baseline models, the LSBAFN obtains a significant improvement of 0.6% and 7.81% in Micro-F1 and Macro-F1 on the Article Topic dataset and improvements of 1.03% and 0.47% in P@k and 1.02% and 0.38% in nDCG@k on the Software Category dataset and RCV1 dataset.

Depression represents a major global health challenge, yet traditional clinical diagnosis faces limitations, including high costs, limited coverage, and low patient willingness. Social media platforms provide new opportunities for early depression screening through user-generated content. However, existing methods often lack systematic integration of clinical knowledge and fail to leverage multi-modal information comprehensively. We propose a DSM-5-guided methodology that systematically maps clinical diagnostic criteria to computable social media features across three modalities: textual semantics (BERT-based deep semantic extraction), behavioral patterns (temporal activity analysis), and topic distributions (LDA-based cognitive bias identification). We design a hierarchical architecture integrating BERT, Bi-LSTM, hierarchical attention, and multi-task learning to capture both character-level and post-level importance while jointly optimizing depression classification, symptom recognition, and severity assessment. Experiments on the WU3D dataset (32,570 users, 2.19 million posts) demonstrate that our model achieves 91.8% F1-score, significantly outperforming baseline methods (BERT: 85.6%, TextCNN: 78.6%, and SVM: 72.1%) and large language models (GPT-4 few-shot: 86.9%). Ablation studies confirm that each component contributes meaningfully with synergistic effects. The model provides interpretable predictions through attention visualization and outputs fine-grained symptom assessments aligned with DSM-5 criteria. With low computational cost (~50 ms inference time), local deployability, and superior privacy protection, our approach offers significant practical value for large-scale mental health screening applications. This work demonstrates that domain-specialized methods with explicit clinical knowledge integration remain highly competitive in the era of general-purpose large language models.

Aspect-level sentiment analysis (ABSA) is a pivotal task within the domain of neurorobotics, contributing to the comprehension of fine-grained textual emotions. Despite the extensive research undertaken on ABSA, the limited availability of training data remains a significant obstacle that hinders the performance of previous studies. Moreover, previous works have predominantly focused on concatenating semantic and syntactic features to predict sentiment polarity, which inadvertently severed the intrinsic connection. Several studies have attempted to utilize multi-layer graph convolution for the purpose of extracting syntactic characteristics. However, this approach has encountered the issue of gradient explosion. This paper investigates the possibilities of leveraging ChatGPT for aspect-level text augmentation. Furthermore, we introduce an improved gated attention mechanism specifically designed for graph convolutional networks to mitigates the problem of gradient explosion. By enriching the features of the dependency graph with a sentiment knowledge base, we strengthen the relationship between aspect words and the polarity of the contextual sentiment. It is worth mentioning that we employ cross-fusion to effectively integrate textual semantic and syntactic features. The experimental results substantiate the superiority of our model over the baseline models in terms of performance.

Text classification is a common way of working in computer applications, in daily work, the accurate and efficient classification of diverse categories of texts Text classification is a common way of working in computer applications, in daily work, the accurate and efficient classification of diverse categories of texts