The task of knowledge graph entity typing (KGET) aims to infer the missing types for entities in knowledge graphs, which is a significant subtask of knowledge graph completion (KGC). In despite of its progress, we observe that the sparsity of the dataset greatly affects the task itself as well as downstream tasks. In this paper, we propose a framework to alleviate this problem, which consists of data augmentation and type inference. We introduce a Statistics-based Entity Type Data Augmentation (SET-DA) method in data augmentation phase, which calculates a type probability distribution for each entity by statistically determining the global relation-type statistics information, and then employ embedding-based KGET models for type inference. Experimental results on two widely used datasets indicate that the proposed framework can solve the data sparsity problem in the task of KGET and the performance of models trained with data augmentation by SET-DA significantly outperforms previous state-of-the-art methods.

Anxiety disorder is a common mental disorder that has received increasing attention due to its high incidence, comorbidity, and recurrence. In recent years, with the rapid development of information technology, social media platforms have become a crucial source of data for studying anxiety disorders. Existing studies on anxiety disorders have focused on utilizing user-generated contents to study correlations with disorders or identify disorders. However, these studies overlook the emotional information in social media posts, restraining the effective capture of users’ emotions or mental states when posting. This article focuses on the sentiment polarity of anxiety-related posts on a Chinese social media and designs sentiment classification models via fuzing linguistic and semantic features of the posts. First, we extract the linguistic features from posts based on the simplified Chinese–Linguistic inquiry and word count (SC-LIWC) dictionary, and propose a novel recursive feature selection algorithm to reserve important linguistic features. Second, we propose a TextCNN-based model to study the deep semantic features of posts and fuze their linguistic features to obtain a better representation. Finally, to conduct anxiety analysis on Chinese social media, we construct a postlevel sentiment analysis dataset based on anxiety-related posts on Sina Weibo. The experimental results indicate that our proposed fusion models exhibit better performance in the task of identifying the sentiment polarity of anxiety-related posts on Chinese social media.

The global prevalence of anxiety disorders is the highest among mental disorders in 2020. However, most people still ignore the danger of anxiety disorders and most of the research on mental disorders only focuses on depression patients. Therefore, this paper makes a Multi-Modal-Anxiety(MMA) dataset for anxiety disorder detection based on data from Weibo social media, and proposes a Multimodal-Anxiety-Detection Network(MADNet) which fused three dimensions: textual information, image information and behavior information. The model maps textual features and non-textual features into the same semantic space for fusion via Multimodal-Anxiety-Information fusion method(MAI) to predict the anxiety tendency for a single post. The experimental results show that the model has achieved F1-score 70.96% and AUC-ROC 70.91% on the MMA dataset, which is state-of-the-art among the existing models. This paper also explores and analyses the prediction of the model through interpretable methods to prove the validity of the model. Overall, this paper provides a usable dataset, model baseline, and multimodal fusion methods for further research on anxiety disorder based on social media. The code associated with this paper is available at https://github.com/Shuzhong-Lai/MADNet.

Depression is a prevalent and severe mental disorder that often goes undetected and untreated, particularly in its early stages. However, social media has emerged as a valuable resource for identifying symptoms of depression and other mental disorders as people are increasingly willing to share their experiences and emotions online. As such, social media-based depression detection has become an important area of research. Unfortunately, despite the growing number of cases in China, there are few Chinese social media-based resources for depression research. To address this gap, this article presents a dataset collected from Sina Weibo and approaches depression detection as a binary classification problem. A depression lexicon is developed based on domain knowledge of depression and the Dalian University of Technology Sentiment Lexicon (DUT-SL), which facilitates better extraction of lexical features related to depression. Then the lexical features are fused using a correlation-based metric. The effectiveness of this approach is verified using five classical machine learning methods and two boosting-based models, both on a public dataset and our dataset. Experimental results indicate that the depression domain lexicon features improve classification performance and fusing these features based on their correlations can further enhance prediction effectiveness. This study provides a method for future research in social media-based depression detection and contributes to the development of Chinese depression detection resources.

Suicide causes serious harm to individuals, families and society, and becomes a social problem of widespread concern. Therefore, it is necessary to find and intervene individuals at risk of suicide as soon as possible. In recent years, social media data has successfully been leveraged for suicide risk detection. However, for fine-grained suicide risk detection, the existing models ignore the deviation between the predicted results and the real results when making wrong predictions, and do not pay attention to the semantic information contained in the labels. This paper proposes a deep learning model based on Label-Text Correlation and Deviation Punishment (LTC-DP). While learning the semantic relation adequately between the text and the corresponding label, the model can give different punishment adaptively according to the deviation degrees between the predicted results and the real result. The experimental results show that compared with the baseline model, the proposed model has better performance in fine-grained suicide risk detection. In addition, we release a fine-grained suicide risk detection data set based on Weibo, the data set is available at https://github.com/cxyazy/FGCSD-main.

Knowledge graphs (KGs) are widely used in various natural language processing applications. In order to expand the coverage of a KG, KG completion has attracted extensive attention. The commonly used embedding methods based on a large amount of training data can play an important role in this work. However, with few of triples, the performance of these methods will be greatly reduced. The completion of this kind of few-shot task is more challenging. In this work, we propose a method of data enhancement to increase the data quantity and solve the problem of sample shortage. Specifically, we first observe that the representation vectors of the relation in a KG are approximately subordinate to Gaussian distribution. Then we construct a Gaussian distribution for the relation of each triple in few-shot task according to the distributions of its similar relations in background graph. Further, we sample from the Gaussian distribution of each triple to expand the training data. Finally, we use an adaptive attentional network model FAAN proposed by Sheng et al. as the baseline model. Experimental results on two public datasets NELL-One and Wiki-One show that the proposed method achieves better performance.

Depression is a common mood disorder, which has the characteristics of high prevalence, high recurrence rate, high disability rate and high mortality rate. There are a large number of medical literature on depression, but the number is large and disorderly, which will undoubtedly increase the burden of biomedical researchers and medical workers to obtain knowledge, and is not conducive to the research on the pathogenesis and treatment of depression. Therefore, we construct a knowledge graph of depression based on biomedical literature to assist the study of depression. We use medical abstracts as the main data source and extract knowledge from them by using SemRep, which is a biomedical information extraction system. Secondly, we use another information extraction tool named OpenIE to correct the data extracted by SemRep. Then, by fusing the extracted knowledge with structured data extracted from SemMedDB, we finally get 8,840 triples which include 3,055 entities and 30 relationships. We store them into the graph database Neo4j to visualize the knowledge graph.

In the past few years, with the rapid growth of digital technologies, Facebook, Twitter, and other social media platforms have become the digital oligarchies, which have the enormous capabilities to potentially control what is discussed in cyberspace. In the digital oligarchy era, social perception and social influence in different complex social systems have evolved quickly. In this article, we conducted large-scale empirical studies on social perception and social influence regarding the Trump phenomenon from personal perception, media, and public attention perspectives. We found that there exist obvious correlations between the posting behavior of Trump and the attention of news media. By constructing public attention networks using complex networks based on Google search information, we further reveal that digital platforms could affect social perception and social influence significantly. Especially, we obtained that the public attention can always be influenced by the political moments.