Objective Glioma is a highly heterogeneous and malignant intracranial tumor that presents challenges for clinical treatment.ELMO domain containing 2(ELMOD2)is a GTPase-activating protein that regulates a range of cell...Objective Glioma is a highly heterogeneous and malignant intracranial tumor that presents challenges for clinical treatment.ELMO domain containing 2(ELMOD2)is a GTPase-activating protein that regulates a range of cellular biological processes.However,its specific role and prognostic value in tumorigenesis are still unknown.This study aimed to assess the prognostic relevance and signaling function of ELMOD2 in gliomas.Methods The Chinese Glioma Genome Atlas(CGGA)and The Cancer Genome Atlas(TCGA)databases were utilized to conduct a comprehensive analysis of the expression profile of ELMOD2 in gliomas,elucidating its associations with clinicopathological parameters and patient prognosis.Single-cell analysis was performed to characterize ELMOD2 expression across distinct glioma cell subpopulations.Gene Ontology(GO),Kyoto Encyclopedia of Genes and Genomes(KEGG)enrichment analyses,and Gene Set Variation Analysis(GSVA)were employed to evaluate the potential biological functions of ELMOD2 in gliomagenesis.Specific small interfering RNAs(siRNAs)were used to knock down ELMOD2 in the glioma cell lines U251 and A172 to assess their cellular behaviors and examine the levels of multiple key signaling molecules associated with the occurrence of gliomas.Results ELMOD2 was overexpressed in gliomas,and this upregulation was correlated with tumor grade,isocitrate dehydrogenase mutation,and 1p/19q codeletion status.Notably,ELMOD2 expression was elevated in classical and mesenchymal subtypes,and single-cell resolution analysis revealed predominant enrichment within malignant cells.Functionally,ELMOD2 regulated cell cycle progression,and its overexpression was related to independent adverse outcomes.In vitro experiments revealed that ELMOD2 was located in the cytoplasm and nucleoplasm.Furthermore,ELMOD2 knockdown reduced proliferation,migration,and invasion and increased apoptosis in U251 and A172 cell lines.Finally,ELMOD2 knockdown significantly decreased p-Erk1/2.Conclusions ELMOD2 expression in glioma is positively correlated with tumorigenesis and is a crucial independent prognostic marker.Thus,ELMOD2 is a promising biomarker and therapeutic target for glioma treatment.展开更多
面向bilibili短视频评论数据的情感分析,旨在挖掘视频观看者对短视频的看法,使视频作者也可以快速得到自己想要的评价,进而对后续作品做出改进。针对短视频评论更新快、词汇新颖、评论过长、一词多义等因素造成的短视频评论情感分析准...面向bilibili短视频评论数据的情感分析,旨在挖掘视频观看者对短视频的看法,使视频作者也可以快速得到自己想要的评价,进而对后续作品做出改进。针对短视频评论更新快、词汇新颖、评论过长、一词多义等因素造成的短视频评论情感分析准确率低的问题,文章构建了bilibili短视频评论数据集,并提出了ELMO(Embedding From Language Model)用以构建动态词向量解决一词多义及新词的问题,通过构建TextCNN和Reformer双通道神经网络结构来提取局部、全局特征。由于Reformer采用了局部敏感哈希的特殊注意力机制,更能联系全局特征,之后将两者得到的结果拼接送入分类器得出情感分析的结果,并将得出的结果与多个深度学习模型进行对比。展开更多
网络欺凌检测是网络空间信息内容安全的重要研究内容,也关乎青少年在线安全.针对目前网络欺凌检测方案存在的训练样本少、难以处理多义词、分类性能不太理想等问题,提出一种ELMo-TextCNN检测模型.该模型首先采用迁移学习思想,利用预训练...网络欺凌检测是网络空间信息内容安全的重要研究内容,也关乎青少年在线安全.针对目前网络欺凌检测方案存在的训练样本少、难以处理多义词、分类性能不太理想等问题,提出一种ELMo-TextCNN检测模型.该模型首先采用迁移学习思想,利用预训练的ELMo(embeddings from language models)生成动态词向量,不仅解决了网络欺凌样本规模小的问题,而且由于ELMo采用了双向长短期记忆(bi-directional long short-term memory,BiLSTM)网络结构,会根据上下文推断每个词对应的词向量,能够根据语境理解多义词.该模型再通过擅长处理短文本数据的TextCNN(text convolutional neural network)提取文本特征,最后经过全连接层输出分类结果.实验结果证明,提出的ELMo-TextCNN检测方法能够处理一词多义,并获得更好的分类检测效果.展开更多
目前情感分析模型通常使用word2vec、GloVe等方法生成静态词向量,并且传统的卷积或循环深度模型无法完整地关注上下文,提取特征不充分,影响情感判断。针对上述问题,提出基于ELMo(embedding from language model)和双向自注意力网络(bidi...目前情感分析模型通常使用word2vec、GloVe等方法生成静态词向量,并且传统的卷积或循环深度模型无法完整地关注上下文,提取特征不充分,影响情感判断。针对上述问题,提出基于ELMo(embedding from language model)和双向自注意力网络(bidirectional self-attention network,Bi-SAN)的中文文本情感分析模型。首先通过ELMo语言模型训练得到融合词语本身和上下文信息的词向量,解决了一词多义的问题;同时使用预训练的skip-gram算法代替随机初始化的ELMo模型的嵌入层,提高模型的收敛速度;之后使用Bi-SAN提取特征,由于自注意力机制,Bi-SAN可以完整地关注每个词的上下文,提取特征更为全面。同现有的多个情感分析模型对比,该模型在酒店评论数据集上和NLPCC2014 task2中文数据集取得了更高的F 1值,验证了模型的有效性。展开更多
挖掘电商评论文本中的电商事件对分析用户购物行为和商品场景分类有重要帮助。该文给出电商事件的定义,将电商事件识别问题转换为序列标注问题,构建了一个基于电商评论文本的电商事件标注数据。该文首先在基于字符的BiLSTM-CRF神经网络...挖掘电商评论文本中的电商事件对分析用户购物行为和商品场景分类有重要帮助。该文给出电商事件的定义,将电商事件识别问题转换为序列标注问题,构建了一个基于电商评论文本的电商事件标注数据。该文首先在基于字符的BiLSTM-CRF神经网络模型上进行扩展,加入语言模型词向量(Embeddings from Language Models,ELMo)来提高识别性能。进而考虑中文字形特征,包括五笔和笔画特征。提出两种引入字形特征的新模型,即在预训练语言模型中结合事件的字形信息进行建模。实验结果表明融入字形特征的ELMo可以进一步提高模型性能。最后,该文分别使用新闻和电商领域两份大规模无标注数据训练语言模型。结果表明,电商领域语料对系统的帮助更大。展开更多
基金supported by grants from the Natural Science Foundation of Guangxi Province(Grant No:2022GXNSFAA035639 and 2023GXNSFBA026092)the National Natural Science Foundation of China(Grant No:81860445 and 82260554)the Innovation Project of Guangxi Graduate Education(Grant No:YCBZ2024118)。
文摘Objective Glioma is a highly heterogeneous and malignant intracranial tumor that presents challenges for clinical treatment.ELMO domain containing 2(ELMOD2)is a GTPase-activating protein that regulates a range of cellular biological processes.However,its specific role and prognostic value in tumorigenesis are still unknown.This study aimed to assess the prognostic relevance and signaling function of ELMOD2 in gliomas.Methods The Chinese Glioma Genome Atlas(CGGA)and The Cancer Genome Atlas(TCGA)databases were utilized to conduct a comprehensive analysis of the expression profile of ELMOD2 in gliomas,elucidating its associations with clinicopathological parameters and patient prognosis.Single-cell analysis was performed to characterize ELMOD2 expression across distinct glioma cell subpopulations.Gene Ontology(GO),Kyoto Encyclopedia of Genes and Genomes(KEGG)enrichment analyses,and Gene Set Variation Analysis(GSVA)were employed to evaluate the potential biological functions of ELMOD2 in gliomagenesis.Specific small interfering RNAs(siRNAs)were used to knock down ELMOD2 in the glioma cell lines U251 and A172 to assess their cellular behaviors and examine the levels of multiple key signaling molecules associated with the occurrence of gliomas.Results ELMOD2 was overexpressed in gliomas,and this upregulation was correlated with tumor grade,isocitrate dehydrogenase mutation,and 1p/19q codeletion status.Notably,ELMOD2 expression was elevated in classical and mesenchymal subtypes,and single-cell resolution analysis revealed predominant enrichment within malignant cells.Functionally,ELMOD2 regulated cell cycle progression,and its overexpression was related to independent adverse outcomes.In vitro experiments revealed that ELMOD2 was located in the cytoplasm and nucleoplasm.Furthermore,ELMOD2 knockdown reduced proliferation,migration,and invasion and increased apoptosis in U251 and A172 cell lines.Finally,ELMOD2 knockdown significantly decreased p-Erk1/2.Conclusions ELMOD2 expression in glioma is positively correlated with tumorigenesis and is a crucial independent prognostic marker.Thus,ELMOD2 is a promising biomarker and therapeutic target for glioma treatment.
文摘面向bilibili短视频评论数据的情感分析,旨在挖掘视频观看者对短视频的看法,使视频作者也可以快速得到自己想要的评价,进而对后续作品做出改进。针对短视频评论更新快、词汇新颖、评论过长、一词多义等因素造成的短视频评论情感分析准确率低的问题,文章构建了bilibili短视频评论数据集,并提出了ELMO(Embedding From Language Model)用以构建动态词向量解决一词多义及新词的问题,通过构建TextCNN和Reformer双通道神经网络结构来提取局部、全局特征。由于Reformer采用了局部敏感哈希的特殊注意力机制,更能联系全局特征,之后将两者得到的结果拼接送入分类器得出情感分析的结果,并将得出的结果与多个深度学习模型进行对比。
文摘网络欺凌检测是网络空间信息内容安全的重要研究内容,也关乎青少年在线安全.针对目前网络欺凌检测方案存在的训练样本少、难以处理多义词、分类性能不太理想等问题,提出一种ELMo-TextCNN检测模型.该模型首先采用迁移学习思想,利用预训练的ELMo(embeddings from language models)生成动态词向量,不仅解决了网络欺凌样本规模小的问题,而且由于ELMo采用了双向长短期记忆(bi-directional long short-term memory,BiLSTM)网络结构,会根据上下文推断每个词对应的词向量,能够根据语境理解多义词.该模型再通过擅长处理短文本数据的TextCNN(text convolutional neural network)提取文本特征,最后经过全连接层输出分类结果.实验结果证明,提出的ELMo-TextCNN检测方法能够处理一词多义,并获得更好的分类检测效果.
文摘目前情感分析模型通常使用word2vec、GloVe等方法生成静态词向量,并且传统的卷积或循环深度模型无法完整地关注上下文,提取特征不充分,影响情感判断。针对上述问题,提出基于ELMo(embedding from language model)和双向自注意力网络(bidirectional self-attention network,Bi-SAN)的中文文本情感分析模型。首先通过ELMo语言模型训练得到融合词语本身和上下文信息的词向量,解决了一词多义的问题;同时使用预训练的skip-gram算法代替随机初始化的ELMo模型的嵌入层,提高模型的收敛速度;之后使用Bi-SAN提取特征,由于自注意力机制,Bi-SAN可以完整地关注每个词的上下文,提取特征更为全面。同现有的多个情感分析模型对比,该模型在酒店评论数据集上和NLPCC2014 task2中文数据集取得了更高的F 1值,验证了模型的有效性。
文摘挖掘电商评论文本中的电商事件对分析用户购物行为和商品场景分类有重要帮助。该文给出电商事件的定义,将电商事件识别问题转换为序列标注问题,构建了一个基于电商评论文本的电商事件标注数据。该文首先在基于字符的BiLSTM-CRF神经网络模型上进行扩展,加入语言模型词向量(Embeddings from Language Models,ELMo)来提高识别性能。进而考虑中文字形特征,包括五笔和笔画特征。提出两种引入字形特征的新模型,即在预训练语言模型中结合事件的字形信息进行建模。实验结果表明融入字形特征的ELMo可以进一步提高模型性能。最后,该文分别使用新闻和电商领域两份大规模无标注数据训练语言模型。结果表明,电商领域语料对系统的帮助更大。
