To improve the accuracy of short text matching,a short text matching method with knowledge and structure enhancement for BERT(KS-BERT)was proposed in this study.This method first introduced external knowledge to the i...To improve the accuracy of short text matching,a short text matching method with knowledge and structure enhancement for BERT(KS-BERT)was proposed in this study.This method first introduced external knowledge to the input text,and then sent the expanded text to both the context encoder BERT and the structure encoder GAT to capture the contextual relationship features and structural features of the input text.Finally,the match was determined based on the fusion result of the two features.Experiment results based on the public datasets BQ_corpus and LCQMC showed that KS-BERT outperforms advanced models such as ERNIE 2.0.This Study showed that knowledge enhancement and structure enhancement are two effective ways to improve BERT in short text matching.In BQ_corpus,ACC was improved by 0.2%and 0.3%,respectively,while in LCQMC,ACC was improved by 0.4%and 0.9%,respectively.展开更多
Modeling and matching texts is a critical issue in natural language processing(NLP) tasks. In order to improve the accuracy of text matching, multi-granularities capture matching features(MG-CMF) model was proposed. T...Modeling and matching texts is a critical issue in natural language processing(NLP) tasks. In order to improve the accuracy of text matching, multi-granularities capture matching features(MG-CMF) model was proposed. The proposed model used convolution operations to construct the representation of text under multiple granularities, used max-pooling operations to filter more reasonable text representations and built a matching matrix at different granularities. Then, the convolution neural network(CNN) was used to capture the matching information in each granularity. Finally, the captured matching features were input into the fully connected neural network to obtain the matching similarity. By making some experiments, the results indicate that the MG-CMF model not only gets multiple granularity representations of sentences but also can obtain matching information from multiple granularities of sentences better than the other text matching models.展开更多
Text classification is an essential task of natural language processing. Preprocessing, which determines the representation of text features, is one of the key steps of text classification architecture. It proposed a ...Text classification is an essential task of natural language processing. Preprocessing, which determines the representation of text features, is one of the key steps of text classification architecture. It proposed a novel efficient and effective preprocessing algorithm with three methods for text classification combining the Orthogonal Matching Pursuit algorithm to perform the classification. The main idea of the novel preprocessing strategy is that it combined stopword removal and/or regular filtering with tokenization and lowercase conversion, which can effectively reduce the feature dimension and improve the text feature matrix quality. Simulation tests on the 20 newsgroups dataset show that compared with the existing state-of-the-art method, the new method reduces the number of features by 19.85%, 34.35%, 26.25% and 38.67%, improves accuracy by 7.36%, 8.8%, 5.71% and 7.73%, and increases the speed of text classification by 17.38%, 25.64%, 23.76% and 33.38% on the four data, respectively.展开更多
Clothing attribute recognition has become an essential technology,which enables users to automatically identify the characteristics of clothes and search for clothing images with similar attributes.However,existing me...Clothing attribute recognition has become an essential technology,which enables users to automatically identify the characteristics of clothes and search for clothing images with similar attributes.However,existing methods cannot recognize newly added attributes and may fail to capture region-level visual features.To address the aforementioned issues,a region-aware fashion contrastive language-image pre-training(RaF-CLIP)model was proposed.This model aligned cropped and segmented images with category and multiple fine-grained attribute texts,achieving the matching of fashion region and corresponding texts through contrastive learning.Clothing retrieval found suitable clothing based on the user-specified clothing categories and attributes,and to further improve the accuracy of retrieval,an attribute-guided composed network(AGCN)as an additional component on RaF-CLIP was introduced,specifically designed for composed image retrieval.This task aimed to modify the reference image based on textual expressions to retrieve the expected target.By adopting a transformer-based bidirectional attention and gating mechanism,it realized the fusion and selection of image features and attribute text features.Experimental results show that the proposed model achieves a mean precision of 0.6633 for attribute recognition tasks and a recall@10(recall@k is defined as the percentage of correct samples appearing in the top k retrieval results)of 39.18 for composed image retrieval task,satisfying user needs for freely searching for clothing through images and texts.展开更多
全球心理健康问题形势严峻,由于心理健康服务的从业人员不足,遭受心理健康困扰的人并不总是能获得专业的心理健康服务.检索式心理健康社区自动问答可以快速地为需要心理健康服务的人提供相应的信息自助服务.与传统检索式社区问答中的文...全球心理健康问题形势严峻,由于心理健康服务的从业人员不足,遭受心理健康困扰的人并不总是能获得专业的心理健康服务.检索式心理健康社区自动问答可以快速地为需要心理健康服务的人提供相应的信息自助服务.与传统检索式社区问答中的文本匹配不同,在匹配支持帖和求助帖时,需要考虑2种不同层面的匹配准则:语义层面和心理层面.为了解决该问题,提出融合角色心理画像的2阶段文本匹配模型(two-stage text matching model integrating characters’mental portrait,T2CMP),该模型引入心理特征用于构建角色心理画像,从而辅助模型理解文本心理层面的内容和匹配关系.同时为了提升检索效率以及减少大量负样例带来的噪声问题,将文本匹配任务拆分为2阶段的序列型子任务.首先针对每条求助帖,使用基于语义的筛选模型甄别出候选支持帖;然后依据用户的角色心理画像,使用多层注意力机制将其与语义信息有效融合,提高模型的总体效果.在MHCQA数据集上的实验结果显示,T2CMP比现有优秀算法拥有更高的F1值.展开更多
在信息检索领域,量子干涉理论已应用于文档相关性、次序效应等核心问题的研究中,旨在建模用户认知引起的类量子干涉现象.文中从语言理解的需求出发,利用量子理论的数学工具分析语义组合过程中存在的语义演化现象,提出融合量子干涉信息...在信息检索领域,量子干涉理论已应用于文档相关性、次序效应等核心问题的研究中,旨在建模用户认知引起的类量子干涉现象.文中从语言理解的需求出发,利用量子理论的数学工具分析语义组合过程中存在的语义演化现象,提出融合量子干涉信息的双重特征文本表示模型(Quantum Interference Based Duet-Feature Text Representation Model,QDTM).模型以约化密度矩阵为语言表示的核心组件,有效建模维度级别的语义干涉信息.在此基础上,构建捕获全局特征信息与局部特征信息的模型结构,满足语言理解过程中不同粒度的语义特征需求.在文本分类数据集和问答数据集上的实验表明,QDTM的性能优于量子启发的语言模型和神经网络文本匹配模型.展开更多
结合胃镜超声和白光内镜可以更准确地识别胃肠道间质瘤.但是现有的多模态方法往往仅关注于图像特征,忽略了诊断文本信息中所包含的语义信息对于精确理解和诊断医学图像的重要性.为此,本文提出一种新的基于文本引导下的多模态医学图像分...结合胃镜超声和白光内镜可以更准确地识别胃肠道间质瘤.但是现有的多模态方法往往仅关注于图像特征,忽略了诊断文本信息中所包含的语义信息对于精确理解和诊断医学图像的重要性.为此,本文提出一种新的基于文本引导下的多模态医学图像分析算法框架(Text-guided Multi-modal Medical image analysis framework,TMM-Net).TMM-Net使用多阶段的诊断文本来引导模型学习,以提取图像中的关键诊断信息特征,然后通过交叉模态注意力机制促进多模态特征之间的交互.值得注意的是,TMM-Net通过预测病变属性来模拟临床诊断过程,从而增强了可解释性.验证实验在两个中心包含10 025个模态数据对的数据集上进行.结果表明,该方法相比目前最优的GISTs诊断方法精度提升7.7%,同时获得了最高的(Area Under the Curve,AUC)值:0.927,其可解释性可以更好地适合临床需求.展开更多
文摘To improve the accuracy of short text matching,a short text matching method with knowledge and structure enhancement for BERT(KS-BERT)was proposed in this study.This method first introduced external knowledge to the input text,and then sent the expanded text to both the context encoder BERT and the structure encoder GAT to capture the contextual relationship features and structural features of the input text.Finally,the match was determined based on the fusion result of the two features.Experiment results based on the public datasets BQ_corpus and LCQMC showed that KS-BERT outperforms advanced models such as ERNIE 2.0.This Study showed that knowledge enhancement and structure enhancement are two effective ways to improve BERT in short text matching.In BQ_corpus,ACC was improved by 0.2%and 0.3%,respectively,while in LCQMC,ACC was improved by 0.4%and 0.9%,respectively.
文摘Modeling and matching texts is a critical issue in natural language processing(NLP) tasks. In order to improve the accuracy of text matching, multi-granularities capture matching features(MG-CMF) model was proposed. The proposed model used convolution operations to construct the representation of text under multiple granularities, used max-pooling operations to filter more reasonable text representations and built a matching matrix at different granularities. Then, the convolution neural network(CNN) was used to capture the matching information in each granularity. Finally, the captured matching features were input into the fully connected neural network to obtain the matching similarity. By making some experiments, the results indicate that the MG-CMF model not only gets multiple granularity representations of sentences but also can obtain matching information from multiple granularities of sentences better than the other text matching models.
文摘Text classification is an essential task of natural language processing. Preprocessing, which determines the representation of text features, is one of the key steps of text classification architecture. It proposed a novel efficient and effective preprocessing algorithm with three methods for text classification combining the Orthogonal Matching Pursuit algorithm to perform the classification. The main idea of the novel preprocessing strategy is that it combined stopword removal and/or regular filtering with tokenization and lowercase conversion, which can effectively reduce the feature dimension and improve the text feature matrix quality. Simulation tests on the 20 newsgroups dataset show that compared with the existing state-of-the-art method, the new method reduces the number of features by 19.85%, 34.35%, 26.25% and 38.67%, improves accuracy by 7.36%, 8.8%, 5.71% and 7.73%, and increases the speed of text classification by 17.38%, 25.64%, 23.76% and 33.38% on the four data, respectively.
基金National Natural Science Foundation of China(No.61971121)。
文摘Clothing attribute recognition has become an essential technology,which enables users to automatically identify the characteristics of clothes and search for clothing images with similar attributes.However,existing methods cannot recognize newly added attributes and may fail to capture region-level visual features.To address the aforementioned issues,a region-aware fashion contrastive language-image pre-training(RaF-CLIP)model was proposed.This model aligned cropped and segmented images with category and multiple fine-grained attribute texts,achieving the matching of fashion region and corresponding texts through contrastive learning.Clothing retrieval found suitable clothing based on the user-specified clothing categories and attributes,and to further improve the accuracy of retrieval,an attribute-guided composed network(AGCN)as an additional component on RaF-CLIP was introduced,specifically designed for composed image retrieval.This task aimed to modify the reference image based on textual expressions to retrieve the expected target.By adopting a transformer-based bidirectional attention and gating mechanism,it realized the fusion and selection of image features and attribute text features.Experimental results show that the proposed model achieves a mean precision of 0.6633 for attribute recognition tasks and a recall@10(recall@k is defined as the percentage of correct samples appearing in the top k retrieval results)of 39.18 for composed image retrieval task,satisfying user needs for freely searching for clothing through images and texts.
文摘全球心理健康问题形势严峻,由于心理健康服务的从业人员不足,遭受心理健康困扰的人并不总是能获得专业的心理健康服务.检索式心理健康社区自动问答可以快速地为需要心理健康服务的人提供相应的信息自助服务.与传统检索式社区问答中的文本匹配不同,在匹配支持帖和求助帖时,需要考虑2种不同层面的匹配准则:语义层面和心理层面.为了解决该问题,提出融合角色心理画像的2阶段文本匹配模型(two-stage text matching model integrating characters’mental portrait,T2CMP),该模型引入心理特征用于构建角色心理画像,从而辅助模型理解文本心理层面的内容和匹配关系.同时为了提升检索效率以及减少大量负样例带来的噪声问题,将文本匹配任务拆分为2阶段的序列型子任务.首先针对每条求助帖,使用基于语义的筛选模型甄别出候选支持帖;然后依据用户的角色心理画像,使用多层注意力机制将其与语义信息有效融合,提高模型的总体效果.在MHCQA数据集上的实验结果显示,T2CMP比现有优秀算法拥有更高的F1值.
文摘在信息检索领域,量子干涉理论已应用于文档相关性、次序效应等核心问题的研究中,旨在建模用户认知引起的类量子干涉现象.文中从语言理解的需求出发,利用量子理论的数学工具分析语义组合过程中存在的语义演化现象,提出融合量子干涉信息的双重特征文本表示模型(Quantum Interference Based Duet-Feature Text Representation Model,QDTM).模型以约化密度矩阵为语言表示的核心组件,有效建模维度级别的语义干涉信息.在此基础上,构建捕获全局特征信息与局部特征信息的模型结构,满足语言理解过程中不同粒度的语义特征需求.在文本分类数据集和问答数据集上的实验表明,QDTM的性能优于量子启发的语言模型和神经网络文本匹配模型.
文摘结合胃镜超声和白光内镜可以更准确地识别胃肠道间质瘤.但是现有的多模态方法往往仅关注于图像特征,忽略了诊断文本信息中所包含的语义信息对于精确理解和诊断医学图像的重要性.为此,本文提出一种新的基于文本引导下的多模态医学图像分析算法框架(Text-guided Multi-modal Medical image analysis framework,TMM-Net).TMM-Net使用多阶段的诊断文本来引导模型学习,以提取图像中的关键诊断信息特征,然后通过交叉模态注意力机制促进多模态特征之间的交互.值得注意的是,TMM-Net通过预测病变属性来模拟临床诊断过程,从而增强了可解释性.验证实验在两个中心包含10 025个模态数据对的数据集上进行.结果表明,该方法相比目前最优的GISTs诊断方法精度提升7.7%,同时获得了最高的(Area Under the Curve,AUC)值:0.927,其可解释性可以更好地适合临床需求.