Word embeddings are effective intermediate representations for capturing semantic regularities between words in natural language processing (NLP) tasks. We propose sentiment-aware word embedding for emotional classification, which consists of integrating sentiment evidence within the emotional embedding component of a term vector. We take advantage of the multiple types of emotional knowledge, just as the existing emotional lexicon, to build emotional word vectors to represent emotional information. Then the emotional word vector is combined with the traditional word embedding to construct the hybrid representation, which contains semantic and emotional information as the inputs of the emotion classification experiments. Our method maintains the interpretability of word embeddings, and leverages external emotional information in addition to input text sequences. Extensive results on several machine learning models show that the proposed methods can improve the accuracy of emotion classification tasks.

One decisive problem of short text classification is the serious dimensional disaster when utilizing a statistics-based approach to construct vector spaces. Here, a feature reduction method is proposed that is based on two-stage feature clustering (TSFC), which is applied to short text classification. Features are semi-loosely clustered by combining spectral clustering with a graph traversal algorithm. Next, intra-cluster feature screening rules are designed to remove outlier feature words, which improves the effect of similar feature clusters. We classify short texts with corresponding similar feature clusters instead of original feature words. Similar feature clusters replace feature words, and the dimension of vector space is significantly reduced. Several classifiers are utilized to evaluate the effectiveness of this method. The results show that the method largely resolves the dimensional disaster and it can significantly improve the accuracy of short text classification.

The spread of misinformation on social media is a serious issue that can have negative consequences for public health and political stability. While detecting and identifying misinformation can be challenging, many attempts have been made to address this problem. However, traditional models that focus on pairwise relationships on misinformation propagation paths may not be effective in capturing the underlying connections among multiple tweets. To address this limitation, the proposed “Conversation-Branch-Tweet” hypergraph convolutional network (CBT-HGCN) uses a hypergraph to represent the internal structure and content of tweet data, with tweets and their replies viewed as nodes and hyperedges, respectively. The model first pre-processes the tweets of a conversation and then uses a pre-trained model as an encoder to extract node information. Finally, a hypergraph convolution network is used as an information fuser for classification. Experimental results on three benchmark datasets (Twitter15, Twitter16, and Pheme) show that the proposed model outperforms several strong baseline models and achieves state-of-the-art performance. This indicates that the CBT-HGCN approach is effective in detecting and identifying misinformation on social media by capturing the underlying connections among multiple tweets.