Audio-Text Multimodal Speech Recognition via Dual-Tower Architecture for Mandarin Air Traffic Control Communications

In air traffic control communications (ATCC), misunderstandings between pilots and controllers could result in fatal aviation accidents. Fortunately, advanced automatic speech recognition technology has emerged as a promising means of preventing miscommunications and enhancing aviation safety. However, most existing speech recognition methods merely incorporate external language models on the decoder side, leading to insufficient semantic alignment between speech and text modalities during the encoding phase. Furthermore, it is challenging to model acoustic context dependencies over long distances due to the longer speech sequences than text, especially for the extended ATCC data. To address these issues, we propose a speech-text multimodal dual-tower architecture for speech recognition. It employs cross-modal interactions to achieve close semantic alignment during the encoding stage and strengthen its capabilities in modeling auditory long-distance context dependencies. In addition, a two-stage training strategy is elaborately devised to derive semantics-aware acoustic representations effectively. The first stage focuses on pre-training the speech-text multimodal encoding module to enhance inter-modal semantic alignment and aural long-distance context dependencies. The second stage fine-tunes the entire network to bridge the input modality variation gap between the training and inference phases and boost generalization performance. Extensive experiments demonstrate the effectiveness of the proposed speech-text multimodal speech recognition method on the ATCC and AISHELL-1 datasets. It reduces the character error rate to 6.54% and 8.73%, respectively, and exhibits substantial performance gains of 28.76% and 23.82% compared with the best baseline model. The case studies indicate that the obtained semantics-aware acoustic representations aid in accurately recognizing terms with similar pronunciations but distinctive semantics. The research provides a novel modeling paradigm for semantics-aware speech recognition in air traffic control communications, which could contribute to the advancement of intelligent and efficient aviation safety management.

基于改进TextRank的科技文本关键词抽取方法

针对科技文本关键词抽取任务中抽取出现次数少但能较好表达文本主旨的词语效果差的问题,提出一种基于改进TextRank的关键词抽取方法。首先,利用词语的词频-逆文档频率(TF-IDF)统计特征和位置特征优化共现图中词语间的概率转移矩阵,通过迭代计算得到词语的初始得分;然后,利用K-Core(K-Core decomposition)算法挖掘KCore子图得到词语的层级特征,利用平均信息熵特征衡量词语的主题表征能力;最后,在词语初始得分的基础上融合层级特征和平均信息熵特征,从而确定关键词。实验结果表明,在公开数据集上,与TextRank方法和OTextRank(Optimized TextRank)方法相比,所提方法在抽取不同关键词数量的实验中,F1均值分别提高了6.5和3.3个百分点;在科技服务项目数据集上,与TextRank方法和OTextRank方法相比,所提方法在抽取不同关键词数量的实验中,F1均值分别提高了7.4和3.2个百分点。实验结果验证了所提方法抽取出现频率低但较好表达文本主旨关键词的有效性。

一种利用词典扩展数据库模式信息的Text2SQL方法

现有Text2SQL方法严重依赖表名和列名在自然语言查询中的显式提及,在同物异名的实际应用场景中准确率急剧下降.此外,这些方法仅仅依赖数据库模式捕捉数据库建模的领域知识,而数据库模式作为结构化的元数据,其表达领域知识的能力是非常有限的,即使有经验的程序员也很难仅从数据库模式完全领会该数据库建模的领域知识,因此程序员必须依赖详细的数据库设计文档才能构造SQL语句以正确地表达特定的查询.为此,本文提出一种利用词典扩展数据库模式信息的Text2SQL方法,该方法从数据库表名和列名解析出其中的单词或短语,查询词典获取这些单词或短语的语义解释,将这些解释看成是相应表名或列名的扩展内容,与表名、列名及其他数据库模式信息(主键、外键等)相结合,作为模型的输入,从而使模型能够更全面地学习数据库建模的应用领域知识.在Spider-syn和Spider数据集上进行的实验说明了所提出方法的有效性,即使自然语言查询中使用的表名和列名与数据库模式中对应的表名和列名完全不同,本文方法也能够得到较好的SQL翻译结果,明显优于最新提出的抗同义词替换攻击的方法.

基于DAN与FastText的藏文短文本分类研究

随着藏文信息不断融入社会生活,越来越多的藏文短文本数据存在网络平台上。针对传统分类方法在藏文短文本上分类性能低的问题,文中提出了一种基于DAN-FastText的藏文短文本分类模型。该模型使用FastText网络在较大规模的藏文语料上进行无监督训练获得预训练的藏文音节向量集,使用预训练的音节向量集将藏文短文本信息转化为音节向量,把音节向量送入DAN(Deep Averaging Networks)网络并在输出阶段融合经过FastText网络训练的句向量特征,最后通过全连接层和softmax层完成分类。在公开的TNCC(Tibetan News Classification Corpus)新闻标题数据集上所提模型的Macro-F1是64.53%,比目前最好评测结果TiBERT模型的Macro-F1得分高出2.81%,比GCN模型的Macro-F1得分高出6.14%,融合模型具有较好的藏文短文本分类效果。

多视图融合DJ-TextRCNN的古籍文本主题推荐研究

传统编目分类和规则匹配方法存在工作效能低、过度依赖专家知识、缺乏对古籍文本自身语义的深层次挖掘、编目主题边界模糊、较难实现对古籍文本领域主题的精准推荐等问题。为此,本文结合古籍语料特征探究如何实现精准推荐符合研究者需求的文本主题内容的方法,以推动数字人文研究的进一步发展。首先,选取本课题组前期标注的古籍语料数据进行主题类别标注和视图分类;其次,构建融合BERT(bidirectional encoder representation from transformers)预训练模型、改进卷积神经网络、循环神经网络和多头注意力机制的语义挖掘模型;最后,融入“主体-关系-客体”多视图的语义增强模型,构建DJ-TextRCNN(DianJi-recurrent convolutional neural networks for text classification)模型实现对典籍文本更细粒度、更深层次、更多维度的语义挖掘。研究结果发现,DJ-TextRCNN模型在不同视图下的古籍主题推荐任务的准确率均为最优。在“主体-关系-客体”视图下,精确率达到88.54%,初步实现了对古籍文本的精准主题推荐,对中华文化深层次、细粒度的语义挖掘具有一定的指导意义。

基于SWPF2vec和DJ-TextRCNN的古籍文本主题分类研究

以编目分类和规则匹配为主的古籍文本主题分类方法存在工作效能低、专家知识依赖性强、分类依据单一化、古籍文本主题自动分类难等问题。对此,本文结合古籍文本内容和文字特征,尝试从古籍内容分类得到符合研究者需求的主题,推动数字人文研究范式的转型。首先,参照东汉古籍《说文解字》对文字的分析方式,以前期标注的古籍语料数据集为基础,构建全新的“字音(说)-原文(文)-结构(解)-字形(字)”四维特征数据集。其次,设计四维特征向量提取模型(speaking,word,pattern,and font to vector,SWPF2vec),并结合预训练模型实现对古籍文本细粒度的特征表示。再其次,构建融合卷积神经网络、循环神经网络和多头注意力机制的古籍文本主题分类模型(dianji-recurrent convolutional neural networks for text classification,DJ-TextRCNN)。最后,融入四维语义特征,实现对古籍文本多维度、深层次、细粒度的语义挖掘。在古籍文本主题分类任务上,DJ-TextRCNN模型在不同维度特征下的主题分类准确率均为最优,在“说文解字”四维特征下达到76.23%的准确率,初步实现了对古籍文本的精准主题分类。

基于BERT和TextCNN的智能制造成熟度评估方法

随着智能制造2025目标的临近,企业为了解自身能力水平纷纷加入到智能制造成熟度评估的行列中。然而,由于智能制造成熟度评估标准的复杂性,企业缺乏其对行业水平的了解,导致企业贸然申请,浪费自身时间的同时又占用大量评估资源。鉴于此,设计了一种新的评估流程,采用文本处理算法对整个评估过程进行了重构,通过利用国标文件中智能制造成熟度评估标准,将其作为训练集,采用基于预训练语言模型与文本神经网络(BERT+TextCNN)相结合的智能评估算法代替人工评估。在真实的企业智能制造数据集上的验证表明,当BERT+TextCNN评估模型在卷积核为[2,3,4]、迭代次数为6次、学习率为3e-5时,对智能制造成熟度进行评估,准确率达到85.32%。这表明所设计的评估方法能够较准确地帮助企业完成智能制造成熟度自评估,有助于企业了解自身智能制造能力水平,制定正确的发展方向。

树立行业发展新方向——Techtextil&Texprocess 2024亮点回顾

为期4天的法兰克福国际产业用纺织品及非织造布展览会及国际纺织品及柔性材料缝制加工展览会(Techtextil&Texprocess 2024)吸引了来自全球53个国家和地区的1700家领先企业参展和来自102个国家和地区的38000名观众,展会规模再创新高,充分彰显了纺织行业蓬勃的生命力与持续的创新力。

Identifying multidisciplinary problems from scientific publications based on a text generation method

Purpose:A text generation based multidisciplinary problem identification method is proposed,which does not rely on a large amount of data annotation.Design/methodology/approach:The proposed method first identifies the research objective types and disciplinary labels of papers using a text classification technique;second,it generates abstractive titles for each paper based on abstract and research objective types using a generative pre-trained language model;third,it extracts problem phrases from generated titles according to regular expression rules;fourth,it creates problem relation networks and identifies the same problems by exploiting a weighted community detection algorithm;finally,it identifies multidisciplinary problems based on the disciplinary labels of papers.Findings:Experiments in the“Carbon Peaking and Carbon Neutrality”field show that the proposed method can effectively identify multidisciplinary research problems.The disciplinary distribution of the identified problems is consistent with our understanding of multidisciplinary collaboration in the field.Research limitations:It is necessary to use the proposed method in other multidisciplinary fields to validate its effectiveness.Practical implications:Multidisciplinary problem identification helps to gather multidisciplinary forces to solve complex real-world problems for the governments,fund valuable multidisciplinary problems for research management authorities,and borrow ideas from other disciplines for researchers.Originality/value:This approach proposes a novel multidisciplinary problem identification method based on text generation,which identifies multidisciplinary problems based on generative abstractive titles of papers without data annotation required by standard sequence labeling techniques.

CVTD: A Robust Car-Mounted Video Text Detector

Text perception is crucial for understanding the semantics of outdoor scenes,making it a key requirement for building intelligent systems for driver assistance or autonomous driving.Text information in car-mounted videos can assist drivers in making decisions.However,Car-mounted video text images pose challenges such as complex backgrounds,small fonts,and the need for real-time detection.We proposed a robust Car-mounted Video Text Detector(CVTD).It is a lightweight text detection model based on ResNet18 for feature extraction,capable of detecting text in arbitrary shapes.Our model efficiently extracted global text positions through the Coordinate Attention Threshold Activation(CATA)and enhanced the representation capability through stacking two Feature Pyramid Enhancement Fusion Modules(FPEFM),strengthening feature representation,and integrating text local features and global position information,reinforcing the representation capability of the CVTD model.The enhanced feature maps,when acted upon by Text Activation Maps(TAM),effectively distinguished text foreground from non-text regions.Additionally,we collected and annotated a dataset containing 2200 images of Car-mounted Video Text(CVT)under various road conditions for training and evaluating our model’s performance.We further tested our model on four other challenging public natural scene text detection benchmark datasets,demonstrating its strong generalization ability and real-time detection speed.This model holds potential for practical applications in real-world scenarios.

From text to image:challenges in integrating vision into ChatGPT for medical image interpretation

Large language models(LLMs),such as ChatGPT developed by OpenAI,represent a significant advancement in artificial intelligence(AI),designed to understand,generate,and interpret human language by analyzing extensive text data.Their potential integration into clinical settings offers a promising avenue that could transform clinical diagnosis and decision-making processes in the future(Thirunavukarasu et al.,2023).This article aims to provide an in-depth analysis of LLMs’current and potential impact on clinical practices.Their ability to generate differential diagnosis lists underscores their potential as invaluable tools in medical practice and education(Hirosawa et al.,2023;Koga et al.,2023).

Method to Remove Handwritten Texts Using Smart Phone

To remove handwritten texts from an image of a document taken by smart phone,an intelligent removal method was proposed that combines dewarping and Fully Convolutional Network with Atrous Convolutional and Atrous Spatial Pyramid Pooling(FCN-AC-ASPP).For a picture taken by a smart phone,firstly,the image is transformed into a regular image by the dewarping algorithm.Secondly,the FCN-AC-ASPP is used to classify printed texts and handwritten texts.Lastly,handwritten texts can be removed by a simple algorithm.Experiments show that the classification accuracy of the FCN-AC-ASPP is better than FCN,DeeplabV3+,FCN-AC.For handwritten texts removal effect,the method of combining dewarping and FCN-AC-ASPP is superior to FCN-AC-ASP alone.

Text-to-SQL文本信息处理技术研究综述

信号与信息处理的需求日益增加,离不开数据处理技术,数据处理需要数据库的支持,然而没有经过训练的使用者会因为不熟悉数据库操作产生诸多问题。文本转结构化查询语言(Text to Structured Query Language,Text-to-SQL)的出现,使用户无需掌握结构化查询语言(Structured Query Language,SQL)也能够熟练操作数据库。介绍Text-to-SQL的研究背景及面临的挑战;介绍Text-to-SQL关键技术、基准数据集、模型演变及最新研究进展,关键技术包括Transformer等主流技术,用于模型训练的基准数据集包括WikiSQL和Spider;介绍Text-to-SQL不同阶段模型的特点,详细阐述Text-to-SQL最新研究成果的工作原理,包括模型构建、解析器设计及数据集生成;总结Text-to-SQL未来的发展方向及研究重点。

Comparing Fine-Tuning, Zero and Few-Shot Strategies with Large Language Models in Hate Speech Detection in English

Large Language Models(LLMs)are increasingly demonstrating their ability to understand natural language and solve complex tasks,especially through text generation.One of the relevant capabilities is contextual learning,which involves the ability to receive instructions in natural language or task demonstrations to generate expected outputs for test instances without the need for additional training or gradient updates.In recent years,the popularity of social networking has provided a medium through which some users can engage in offensive and harmful online behavior.In this study,we investigate the ability of different LLMs,ranging from zero-shot and few-shot learning to fine-tuning.Our experiments show that LLMs can identify sexist and hateful online texts using zero-shot and few-shot approaches through information retrieval.Furthermore,it is found that the encoder-decoder model called Zephyr achieves the best results with the fine-tuning approach,scoring 86.811%on the Explainable Detection of Online Sexism(EDOS)test-set and 57.453%on the Multilingual Detection of Hate Speech Against Immigrants and Women in Twitter(HatEval)test-set.Finally,it is confirmed that the evaluated models perform well in hate text detection,as they beat the best result in the HatEval task leaderboard.The error analysis shows that contextual learning had difficulty distinguishing between types of hate speech and figurative language.However,the fine-tuned approach tends to produce many false positives.

Introducing MagBERT:A language model for magnesium textual data mining and analysis

Magnesium(Mg)based materials hold immense potential for various applications due to their lightweight and high strength-to-weight ratio.However,to fully harness the potential of Mg alloys,structured analytics are essential to gain valuable insights from centuries of accumulated knowledge.Efficient information extraction from the vast corpus of scientific literature is crucial for this purpose.In this work,we introduce MagBERT,a BERT-based language model specifically trained for Mg-based materials.Utilizing a dataset of approximately 370,000 abstracts focused on Mg and its alloys,MagBERT is designed to understand the intricate details and specialized terminology of this domain.Through rigorous evaluation,we demonstrate the effectiveness of MagBERT for information extraction using a fine-tuned named entity recognition(NER)model,named MagNER.This NER model can extract mechanical,microstructural,and processing properties related to Mg alloys.For instance,we have created an Mg alloy dataset that includes properties such as ductility,yield strength,and ultimate tensile strength(UTS),along with standard alloy names.The introduction of MagBERT is a novel advancement in the development of Mg-specific language models,marking a significant milestone in the discovery of Mg alloys and textual information extraction.By making the pre-trained weights of MagBERT publicly accessible,we aim to accelerate research and innovation in the field of Mg-based materials through efficient information extraction and knowledge discovery.

YOLOv5ST:A Lightweight and Fast Scene Text Detector

Scene text detection is an important task in computer vision.In this paper,we present YOLOv5 Scene Text(YOLOv5ST),an optimized architecture based on YOLOv5 v6.0 tailored for fast scene text detection.Our primary goal is to enhance inference speed without sacrificing significant detection accuracy,thereby enabling robust performance on resource-constrained devices like drones,closed-circuit television cameras,and other embedded systems.To achieve this,we propose key modifications to the network architecture to lighten the original backbone and improve feature aggregation,including replacing standard convolution with depth-wise convolution,adopting the C2 sequence module in place of C3,employing Spatial Pyramid Pooling Global(SPPG)instead of Spatial Pyramid Pooling Fast(SPPF)and integrating Bi-directional Feature Pyramid Network(BiFPN)into the neck.Experimental results demonstrate a remarkable 26%improvement in inference speed compared to the baseline,with only marginal reductions of 1.6%and 4.2%in mean average precision(mAP)at the intersection over union(IoU)thresholds of 0.5 and 0.5:0.95,respectively.Our work represents a significant advancement in scene text detection,striking a balance between speed and accuracy,making it well-suited for performance-constrained environments.

Multi-Objective Equilibrium Optimizer for Feature Selection in High-Dimensional English Speech Emotion Recognition

Speech emotion recognition(SER)uses acoustic analysis to find features for emotion recognition and examines variations in voice that are caused by emotions.The number of features acquired with acoustic analysis is extremely high,so we introduce a hybrid filter-wrapper feature selection algorithm based on an improved equilibrium optimizer for constructing an emotion recognition system.The proposed algorithm implements multi-objective emotion recognition with the minimum number of selected features and maximum accuracy.First,we use the information gain and Fisher Score to sort the features extracted from signals.Then,we employ a multi-objective ranking method to evaluate these features and assign different importance to them.Features with high rankings have a large probability of being selected.Finally,we propose a repair strategy to address the problem of duplicate solutions in multi-objective feature selection,which can improve the diversity of solutions and avoid falling into local traps.Using random forest and K-nearest neighbor classifiers,four English speech emotion datasets are employed to test the proposed algorithm(MBEO)as well as other multi-objective emotion identification techniques.The results illustrate that it performs well in inverted generational distance,hypervolume,Pareto solutions,and execution time,and MBEO is appropriate for high-dimensional English SER.

Assessing trends in wildland-urban interface fire research through text mining: a comprehensive analysis of published literature

Research on fires at the wildland-urban inter-face(WUI)has generated significant insights and advance-ments across various fields of study.Environmental,agri-culture,and social sciences have played prominent roles in understanding the impacts of fires in the environment,in protecting communities,and addressing management challenges.This study aimed to create a database using a text mining technique for global researchers interested in WUI-projects and highlighting the interest of countries in this field.Author’s-Keywords analysis emphasized the dominance of fire science-related terms,especially related to WUI,and identified keyword clusters related to the WUI fire-risk-assessment-system-“exposure”,“danger”,and“vulnerability”within wildfire research.Trends over the past decade showcase shifting research interests with a growing focus on WUI fires,while regional variations highlighted that the“exposure”keyword cluster received greater atten-tion in the southern Europe and South America.However,vulnerability keywords have relatively a lower representation across all regions.The analysis underscores the interdisci-plinary nature of WUI research and emphasizes the need for targeted approaches to address the unique challenges of the wildland-urban interface.Overall,this study provides valu-able insights for researchers and serves as a foundation for further collaboration in this field through the understanding of the trends over recent years and in different regions.

Relational Turkish Text Classification Using Distant Supervised Entities and Relations

Text classification,by automatically categorizing texts,is one of the foundational elements of natural language processing applications.This study investigates how text classification performance can be improved through the integration of entity-relation information obtained from the Wikidata(Wikipedia database)database and BERTbased pre-trained Named Entity Recognition(NER)models.Focusing on a significant challenge in the field of natural language processing(NLP),the research evaluates the potential of using entity and relational information to extract deeper meaning from texts.The adopted methodology encompasses a comprehensive approach that includes text preprocessing,entity detection,and the integration of relational information.Experiments conducted on text datasets in both Turkish and English assess the performance of various classification algorithms,such as Support Vector Machine,Logistic Regression,Deep Neural Network,and Convolutional Neural Network.The results indicate that the integration of entity-relation information can significantly enhance algorithmperformance in text classification tasks and offer new perspectives for information extraction and semantic analysis in NLP applications.Contributions of this work include the utilization of distant supervised entity-relation information in Turkish text classification,the development of a Turkish relational text classification approach,and the creation of a relational database.By demonstrating potential performance improvements through the integration of distant supervised entity-relation information into Turkish text classification,this research aims to support the effectiveness of text-based artificial intelligence(AI)tools.Additionally,it makes significant contributions to the development ofmultilingual text classification systems by adding deeper meaning to text content,thereby providing a valuable addition to current NLP studies and setting an important reference point for future research.

Leveraging Uncertainty for Depth-Aware Hierarchical Text Classification

Hierarchical Text Classification(HTC)aims to match text to hierarchical labels.Existing methods overlook two critical issues:first,some texts cannot be fully matched to leaf node labels and need to be classified to the correct parent node instead of treating leaf nodes as the final classification target.Second,error propagation occurs when a misclassification at a parent node propagates down the hierarchy,ultimately leading to inaccurate predictions at the leaf nodes.To address these limitations,we propose an uncertainty-guided HTC depth-aware model called DepthMatch.Specifically,we design an early stopping strategy with uncertainty to identify incomplete matching between text and labels,classifying them into the corresponding parent node labels.This approach allows us to dynamically determine the classification depth by leveraging evidence to quantify and accumulate uncertainty.Experimental results show that the proposed DepthMatch outperforms recent strong baselines on four commonly used public datasets:WOS(Web of Science),RCV1-V2(Reuters Corpus Volume I),AAPD(Arxiv Academic Paper Dataset),and BGC.Notably,on the BGC dataset,it improvesMicro-F1 andMacro-F1 scores by at least 1.09%and 1.74%,respectively.