RB-GAT: A Text Classification Model Based on RoBERTa-BiGRU with Graph ATtention Network.

With the development of deep learning, several graph neural network (GNN)-based approaches have been utilized for text classification. However, GNNs encounter challenges when capturing contextual text information within a document sequence. To address this, a novel text classification model, RB-GAT, is proposed by combining RoBERTa-BiGRU embedding and a multi-head Graph ATtention Network (GAT). First, the pre-trained RoBERTa model is exploited to learn word and text embeddings in different contexts. Second, the Bidirectional Gated Recurrent Unit (BiGRU) is employed to capture long-term dependencies and bidirectional sentence information from the text context. Next, the multi-head graph attention network is applied to analyze this information, which serves as a node feature for the document. Finally, the classification results are generated through a Softmax layer. Experimental results on five benchmark datasets demonstrate that our method can achieve an accuracy of 71.48%, 98.45%, 80.32%, 90.84%, and 95.67% on Ohsumed, R8, MR, 20NG and R52, respectively, which is superior to the existing nine text classification approaches.

Concurrent and longitudinal contributions of phonological awareness to early adolescent Chinese reading acquisition.

This study investigated universal and language-specific phonological awareness in reading development among Chinese early adolescent students. Seventy-six children participated in this study and completed a series of reading tasks at two data collection points across Grades 5 and 6. In Grade 5, universal phonological awareness (syllable, onset, rhyme, and phoneme awareness), language-specific phonological awareness (tone awareness) as well as character recognition and production measurements were administered to the participants. Lexical inferencing ability was measured in Grade 6. Character recognition and lexical inference were coded as the outcome variables. Subsequent multiple regression analyses showed that Time 1 (Grade 5) language-universal onset and phoneme awareness predicted character recognition and production at Time 1. More strikingly, the study demonstrated that language-specific tone awareness exerted a longitudinal effect on later lexical inferencing ability after controlling for age and nonverbal intelligence. Results underscored both the universality and language specificity of phonological awareness and provided empirical evidence to substantiate the facilitative role of early language-specific psycholinguistic grain size in later reading performance.

Word Knowledge in L2 Chinese Lexical Inference: A Moderated Path Analysis of Language Proficiency Level and Heritage Status.

This study explored the effect of word knowledge facets (word-general and word-specific knowledge) on second language (L2) Chinese lexical inference by highlighting the moderating effect of language proficiency level and learners' heritage status. L2 Chinese learners with a mixture of linguistic (low-intermediate and high-intermediate) and cultural (heritage and non-heritage) backgrounds completed a series of word-knowledge measurements as well as a lexical inferencing task. Through a moderated path model, the study demonstrated that word-general knowledge (morphological awareness) and word-specific knowledge (vocabulary knowledge) contributed to L2 Chinese lexical inference. In addition, the study underlined the moderating effect of heritage status on the correlation between word knowledge and lexical inference. Given the distinct patterns between heritage and non-heritage learners, morphological awareness may define the characteristics of reading profiles in the Chinese heritage learner population.

Regulation of the NK-1 receptor gene expression in human macrophage cells via an NF-kappa B site on its promoter.

We report here that human monocytic/macrophage THP-1 cells express the neurokinin 1 receptor (NK-1R), and that exposure of these cells to the proinflammatory cytokine IL-1 beta increased the expression of the NK-1R gene at the mRNA and protein levels. Because IL-1 beta function involves nuclear factor kappa B (NF-kappa B) activation, these data suggest that this increase in the expression of the NK-1R gene is mediated by the NF-kappa B transcription factor. An earlier report noted that the promoter region of the human NK-1R gene contains a putative binding site for NF-kappa B [Takahashi, K., Tanaka, A., Hara, M. & Nakanishi, S. (1992) Eur. J. Biochem. 204, 1025-1033]. Here we demonstrate that this is indeed a functional NF-kappa B-binding site, and that NF-kappa B is responsible for regulating the expression of the NK-1R gene by binding to the promoter region of the NK-1R gene. To further substantiate that the observed NF-kappa B-dependent IL-1 beta induction of the human NK-1R gene is regulated via a transcriptional event through this NF-kappa B site on the NK-1R gene promoter, we transfected THP-1 cells with a luciferase promoter-reporter construct containing the 5' promoter region of the human NK-1R gene. Exposure of these cells to IL-1 beta or overexpression of NF-kappa B cDNAs resulted in a significant increase in the amount of luciferase activity that was diminished greatly in cells transfected with I kappa B alpha, the NF-kappa B inhibitor. These results directly implicate NF-kappa B in the regulation of the NK-1R gene and provide a molecular mechanism for the increase in expression of the NK-1R gene in responsive cells.