Research and Enlightenment of Text Mining Applications in ADR from Social Media

Objective To discuss how to use social media data for post-marketing drug safety monitoring in China as soon as possible by systematically combing the text mining applications,and to provide new ideas and methods for pharmacovigilance.Methods Relevant domestic and foreign literature was used to explore text classification based on machine learning,text mining based on deep learning(neural networks)and adverse drug reaction(ADR)terminology.Results and Conclusion Text classification based on traditional machine learning mainly include support vector machine(SVM)algorithm,naive Bayesian(NB)classifier,decision tree,hidden Markov model(HMM)and bidirectional en-coder representations from transformers(BERT).The main neural network text mining based on deep learning are convolution neural network(CNN),recurrent neural network(RNN)and long short-term memory(LSTM).ADR terminology standardization tools mainly include“Medical Dictionary for Regulatory Activities”(MedDRA),“WHODrug”and“Systematized Nomenclature of Medicine-Clinical Terms”(SNOMED CT).

An Efficient Character-Level Adversarial Attack Inspired by Textual Variations in Online Social Media Platforms

In recent years,the growing popularity of social media platforms has led to several interesting natural language processing(NLP)applications.However,these social media-based NLP applications are subject to different types of adversarial attacks due to the vulnerabilities of machine learning(ML)and NLP techniques.This work presents a new low-level adversarial attack recipe inspired by textual variations in online social media communication.These variations are generated to convey the message using out-of-vocabulary words based on visual and phonetic similarities of characters and words in the shortest possible form.The intuition of the proposed scheme is to generate adversarial examples influenced by human cognition in text generation on social media platforms while preserving human robustness in text understanding with the fewest possible perturbations.The intentional textual variations introduced by users in online communication motivate us to replicate such trends in attacking text to see the effects of such widely used textual variations on the deep learning classifiers.In this work,the four most commonly used textual variations are chosen to generate adversarial examples.Moreover,this article introduced a word importance ranking-based beam search algorithm as a searching method for the best possible perturbation selection.The effectiveness of the proposed adversarial attacks has been demonstrated on four benchmark datasets in an extensive experimental setup.

Social Media Cyberbullying Detection on Political Violence from Bangla Texts Using Machine Learning Algorithm

When someone threatens or humiliates another person online by sending those unpleasant messages or comments, this is known as Cyberbullying. Recently, Bangla text has been used much more often on social media. People communicate with others on social media through messages and comments. So bullies use social media as a rich environment to bully others, especially on political issues. Fights over Cyberbullying on political and social media posts are common today. Most of the time, it does a lot of damage. However, few works have been done for monitoring Bangla text on social media & no work has been done yet for detecting the bullying Bangla text on political issues due to the lack of annotated corpora and morphologic analyzers. In this work, we used several machine learning classifiers & a model. That will help to detect the Bangla bullying texts on social media. For this work, 11,000 Bangla texts have been collected from the comments section of political Facebook posts to make a new dataset and labelled the data as either bullied or not. This dataset has been used to train the machine learning classifier. The results indicate that Random Forest achieves superior accuracy of 91.08%.

Analyzing COVID-19 Discourse on Twitter: Text Clustering and Classification Models for Public Health Surveillance

Social media has revolutionized the dissemination of real-life information,serving as a robust platform for sharing life events.Twitter,characterized by its brevity and continuous flow of posts,has emerged as a crucial source for public health surveillance,offering valuable insights into public reactions during the COVID-19 pandemic.This study aims to leverage a range of machine learning techniques to extract pivotal themes and facilitate text classification on a dataset of COVID-19 outbreak-related tweets.Diverse topic modeling approaches have been employed to extract pertinent themes and subsequently form a dataset for training text classification models.An assessment of coherence metrics revealed that the Gibbs Sampling Dirichlet Mixture Model(GSDMM),which utilizes trigram and bag-of-words(BOW)feature extraction,outperformed Non-negative Matrix Factorization(NMF),Latent Dirichlet Allocation(LDA),and a hybrid strategy involving Bidirectional Encoder Representations from Transformers(BERT)combined with LDA and K-means to pinpoint significant themes within the dataset.Among the models assessed for text clustering,the utilization of LDA,either as a clustering model or for feature extraction combined with BERT for K-means,resulted in higher coherence scores,consistent with human ratings,signifying their efficacy.In particular,LDA,notably in conjunction with trigram representation and BOW,demonstrated superior performance.This underscores the suitability of LDA for conducting topic modeling,given its proficiency in capturing intricate textual relationships.In the context of text classification,models such as Linear Support Vector Classification(LSVC),Long Short-Term Memory(LSTM),Bidirectional Long Short-Term Memory(BiLSTM),Convolutional Neural Network with BiLSTM(CNN-BiLSTM),and BERT have shown outstanding performance,achieving accuracy and weighted F1-Score scores exceeding 80%.These results significantly surpassed other models,such as Multinomial Naive Bayes(MNB),Linear Support Vector Machine(LSVM),and Logistic Regression(LR),which achieved scores in the range of 60 to 70 percent.

Effective short text classification via the fusion of hybrid features for IoT social data

Nowadays short texts can be widely found in various social data in relation to the 5G-enabled Internet of Things (IoT). Short text classification is a challenging task due to its sparsity and the lack of context. Previous studies mainly tackle these problems by enhancing the semantic information or the statistical information individually. However, the improvement achieved by a single type of information is limited, while fusing various information may help to improve the classification accuracy more effectively. To fuse various information for short text classification, this article proposes a feature fusion method that integrates the statistical feature and the comprehensive semantic feature together by using the weighting mechanism and deep learning models. In the proposed method, we apply Bidirectional Encoder Representations from Transformers (BERT) to generate word vectors on the sentence level automatically, and then obtain the statistical feature, the local semantic feature and the overall semantic feature using Term Frequency-Inverse Document Frequency (TF-IDF) weighting approach, Convolutional Neural Network (CNN) and Bidirectional Gate Recurrent Unit (BiGRU). Then, the fusion feature is accordingly obtained for classification. Experiments are conducted on five popular short text classification datasets and a 5G-enabled IoT social dataset and the results show that our proposed method effectively improves the classification performance.

Research on Feature Extraction Method of Social Network Text

The development of various applications based on social network text is in full swing.Studying text features and classifications is of great value to extract important information.This paper mainly introduces the common feature selection algorithms and feature representation methods,and introduces the basic principles,advantages and disadvantages of SVM and KNN,and the evaluation indexes of classification algorithms.In the aspect of mutual information feature selection function,it describes its processing flow,shortcomings and optimization improvements.In view of its weakness in not balancing the positive and negative correlation characteristics,a balance weight attribute factor and feature difference factor are introduced to make up for its deficiency.The experimental stage mainly describes the specific process:the word segmentation processing,to disuse words,using various feature selection algorithms,including optimized mutual information,and weighted with TF-IDF.Under the two classification algorithms of SVM and KNN,we compare the merits and demerits of all the feature selection algorithms according to the evaluation index.Experiments show that the optimized mutual information feature selection has good performance and is better than KNN under the SVM classification algorithm.This experiment proves its validity.

The Effect of Latinization on Reading Time and Understanding: Greeklish in Communication and Social Media

The increasing prevalence of technology in society has an impact on young people’s language use and development. Greeklish is the writing of Greek texts using the Latin instead of the Greek alphabet, a practice known as Latinization, also employed for many non-latin alphabet languages. The primary aim of this research is to evaluate the effect of Greeklish on reading time. A sample of 732 young Greeks were asked about their habits when communicating through e-mail and social media with their friends and they then participated in an experiment in which they were asked to read and understand two short texts, one written in Greek and the other in Greeklish. The findings of the research show that nearly one third of the participants use Greeklish. The results of the experiment conducted reveal that understanding is not affected by the alphabet used but reading Greeklish is significantly more time consuming than reading Greek independently of the sex and the familiarity of the participants with Greeklish. The findings suggest that amending social and communication media with software utilities related to Latinization such as language identifiers and converters may reduce reading time and thus facilitate written communication among the users.

Artificial Intelligence and the Sustainable Development Goals: An Exploratory Study in the Context of the Society Domain

Artificial Intelligence (AI) experienced significant advancements in recent years, and its potential power is already recognized across various industries. Yet, the rise of AI has led to a growing concern about its impact on meeting the Sustainable Development Goals (SDGs). The aim of this paper was to evaluate contributions and the potential influence of AI to sustainable development in the society domain. Furthermore, the study analyzed GPT-3 responses, as one of the largest language models developed by OpenAI, descriptively. We conducted a set of queries on the SDGs to gather information on GPT-3’s perceptions of AI impact on sustainable development. Analysis of GPT-3’s contribution potential towards the SDGs showcased its broad range of capabilities for contributing to the SDGs in areas such as education, health, and communication. The study findings provide valuable insights into the contributions of AI to sustainable development in the society domain and highlight the importance of proper regulations to promote the responsible use of AI for sustainable development. We highlighted the potential for improvement in neural language processing skills of GPT-3 by avoiding imitating weak human writing styles with more mistakes in longer texts.

谱聚类和Apriori算法在建筑坍塌事故致因组合分析中的应用

建筑坍塌事故是人员伤亡和经济损失较大的事故类型之一。为探究建筑坍塌事故不同致因之间的关联和相互依存关系,首先,选取国内2015—2020年231份建筑坍塌事故报告作为研究对象,借助R语言平台进行文本挖掘,得到43个致因。其次,运用Python进行谱聚类,根据致因之间的关联强度对其进行聚类。最后,利用关联规则挖掘Apriori算法确定建筑坍塌事故致因之间的关键关联组合。结果表明,43个事故致因可分为5类,在每一个簇类中确定了最关键的致因组合,并提出了针对性的预防措施,为坍塌事故的预防和控制提供一种新的思路。

“双减”政策何以有效落地?基于“任务-主体-结构”的政策文本分析和对S省的调查

聚焦“双减”政策何以有效落地问题进行了两个新尝试。一是在研究范式上,从应然逻辑、实然状态和必然进路三个层面回应。二是在分析框架上,融合政策执行过程相关研究和结构化理论构建了突出目标导向、强化结构分析的“任务-主体-结构”政策执行过程模型,更契合我国复杂的政策执行环境。基于对“双减”政策文本和S省的落地实践分析发现:第一阶段的落地过程呈现出“运动式”治理的特征,但实际执行结构、落地条件和落地成效在内涵上远未达到设定目标。建议着力推动向“动员式”转型,以高质量教育体系建设为总抓手,有效回应家长对政策的价值需求。

基于社交媒体数据的城市洪涝灾害信息智能提取与分析

近年来,由于气候变化导致极端降雨引起的城市内涝灾害事件频发,给我国城市水安全和可持续发展带来威胁,准确掌握受灾区域的舆论主体和公众情绪,对提高应急管理部门内涝灾害的态势感知能力具有重要意义。在当今智能网络时代,人们通过社交媒体反映问题和建议的诉求日益凸显,社交媒体已逐渐成为反映民众情感和社会舆情的主要载体,为获取自然灾害信息提供了新的途径。如何从社交媒体中快速提取城市洪涝灾害信息,并对自然灾害信息进行主题分类和情感分析,准确掌握区域灾情的主题类别和民众舆论倾向,是目前亟待解决的关键技术问题。以新浪微博为例,阐述了洪涝灾害数据的获取与预处理方法,构建了基于FastText的城市洪涝灾害信息主题分类和情感分析模型,以准确掌握受灾区域的主题类别和舆论导向。以2021年郑州“7.20”特大暴雨期间洪涝灾害为例的研究结果表明,本文方法实现了对社交媒体中城市洪涝灾害数据的智能提取与分析,主题分类模型对预设八种类别数据的分类预测F1值达到0.80以上,且情感分析模型基本能够准确预测情感标记为“负面”的数据,这表明本文构建的基于FastText的城市洪涝灾害信息主题分类和情感分析模型能够满足支撑城市应急管理部门动态掌握洪涝灾害发展态势及公众情绪的需求,对防涝减灾调度、安抚民众情绪和实时定点救援等工作具有重要的指导意义。

中国革命与中国调查:以太行根据地调查文本和实践为考察中心

抗战全面爆发后中国共产党东渡黄河开辟华北根据地,太行是其中重要的一块。面对复杂的政治局面、残酷的战争环境和广大农村的地域传统,中共结合形势发展及政权建设的需要,开展了规模化的调查研究。通过对支部发展状况和群众动员的调查,中共有针对性地作出适时调整,极大地推进了支部的建立巩固,也形塑了良好的党群关系。正是经过战争年代的斗争洗礼,调查研究不仅成为中共普遍性的工作方法,而且因其始终贯穿于根据地的党政建设和社会改造,逐步发展成为具有独特政党色彩的原则理念及社会治理范式。

基于社交媒体文本挖掘的居民低碳出行意向分析

城市交通运输是碳减排的重要领域,其中城市居民出行碳排放占比达到了20%,低碳出行对缓解全球气候变化具有重要意义。了解居民对低碳出行的意向有助于推广该行为,社交媒体平台提供了大量有价值的信息,文章基于新浪微博中的低碳出行博文数据,采用BERT-BiLSTM模型、LDA主题模型的文本挖掘方法分析居民对低碳出行的行为意向和关注主题。结果表明:居民对低碳出行整体上持积极意向;地铁和公交最受欢迎;低碳出行意向是不同因素综合作用的结果;明星效应对低碳出行意向影响显著。研究结论将有助于低碳出行政策的完善。

顾客“故事讲述”与社会认知的交互——来自经济型酒店在线评论的证据

随着互联网技术的发展,人们越来越依赖在线数据做出决策。然而,很少有研究通过对在线评论数据的挖掘,探讨经济型酒店的顾客在消费中呈现的个体体验与社会认知的交互过程及结果。本文通过运用网络文本分析方法,可视化地对经济型酒店的社会认知与价值感知展开探讨,并运用扎根理论予以归纳总结,揭示经济型酒店的社会认知与个体价值感知的交互过程,指出交互结果基本内容。研究发现经济型酒店的社会认知主要关联于酒店内部匹配顾客基本住宿需求状况,价值感知体现出了强烈的实用主义特征,积极作用得到较高社会认可。服务标准化与环境治理是经济型酒店获得顾客积极情感表达的基础,品质体验则是加分项。通过对真实评论进行分析,精细化提炼顾客关于经济型酒店价值功能的社会共识性元素,对于企业价值创新管理有重要意义,并为接待业在特定细分市场的运作提供了理论依据。

中国省域新质生产力空间网络结构动态演进及驱动力分析

为推动区域协调发展和新质生产力的均衡布局,本文基于K-means聚类分析和随机森林算法测算了2012—2022年中国30个省份的新质生产力发展水平,采用社会网络分析方法系统地研究了中国省域新质生产力空间网络的动态演进特征,并结合文本分析和QAP回归模型探讨了省域新质生产力空间网络差异的驱动因素。研究发现:中国新质生产力整体上呈上升趋势,但区域间发展不均衡问题突出,呈现出东强西弱的特点;在空间分布上,省域新质生产力空间网络的复杂度逐年增加,网络关联和互动不断增强,其中东部沿海省份始终处于核心区域,中部省份逐渐进入核心区,而东北地区始终处于边缘区域。此外,QAP回归结果显示,加强技术进步、提高人力资本素质和有效利用数据要素可以显著提升省域新质生产力水平,推动区域经济的协调和可持续发展。

基于机器学习的自杀意念原因特征分析

自杀是世界上最重大的公共卫生危机之一,它已超过战争、他杀和自然灾害加在一起的死亡总和。本文在具有自杀意念的社交媒体的文本中使用计算机技术、机器学习和深度学习的方法,自动抽取自杀意念原因,并探索内容特征(词、词性、语法)和情感心理特征(语言、情感、自杀心理)对自杀意念原因自动抽取任务的影响。实验结果表明,内容特征作为特征中最主要和最重要的特征表现较好,其中词特征的表现最好,而词性特征和语法特征由于词本身的包含关系,在某种程度上被词特征所覆盖。情感心理特征则对内容特征有较好的完善和补充的效果,情感、情绪或心理的表达对自杀意念原因有较相关的正比例关系。

基于社会网络分析法的建筑施工高处坠落事故成因研究——以华东地区为例

明确建筑施工高处坠落事故发生的关键成因及其关联性,可以有效预防事故的发生,为此类事故的防控工作提供参考。文章收集2012—2021年间华东地区的145份建筑施工高处坠落事故调查报告,基于文本挖掘进行结构化处理,识别出25项成因要素,运用社会网络分析法,借助社会网络分析软件Ucinet构建建筑施工高处坠落事故成因关系网络,分析了整体网络密度、个体网络密度、网络中心性以及核心边缘结构。结果表明:主体责任不落实、防护措施不到位、监督检查不力、现场管理不到位、安全教育培训不到位、安全意识淡薄、防护用品使用不当等7项要素是高处坠落事故中的关键成因,可将其作为建筑施工高处作业的安全防控重点,以减少高处坠落事故的发生。

基于CNN-LSTM的社交媒体大数据评论文本情感元自动识别方法

为了准确识别社交媒体评论文本情感,助力公众负面情绪引导,提出了基于CNN-LSTM的社交媒体大数据评论文本情感元自动识别方法。通过社交媒体大数据分类,并通过具有字典功能的Token将评论文本转换成数字列表。结合词嵌入技术得到向量列表,完成社交媒体大数据向量转换的预处理。将预处理获取的向量列表输入CNN网络,得到评论文本情感元最终局部特征值。将该值传至LSTM,通过遗忘门、输入门、输出门调节,获取评论文本情感元特征表征结果,经Softmax分类器分类后,实现情感元自动识别。实验结果表明,该方法能有效完成实验数据预处理,用文字和标签的形式标记正面、负面情感元,并准确识别情感元,间接反映社会问题,应用性较强。

村落里的故事,故事里的村落:文化图式理论视角下《村落的终结》英译本中文化负载词的英译探析

文化负载词在中国文化对外传播的过程中发挥着举足轻重的作用,由于中西方文化底蕴、思维方式的不同,使得西方读者对文化负载词的认知存在偏差,其在语言转换的过程中难度较大,因此研究文化负载词的英译策略具有重要的理论以及现实意义。在含有较多文化负载词的作品中,目标语读者对其存在理解上的难题,文化图式理论的运用可帮助译者对译语文化负载词进行解码,再进行重新编码,最后传递给目标受众。本文拟将文化图式理论作为分析视角,选取社科类文本《村落的终结》中的文化负载词英译的实例,旨在探讨翻译过程中存在的对应、冲突、缺省这三类文化图式情况的翻译策略,并总结相应的经验,以期推广至以后的文化负载词英译中。

新业态从业人员社会保障政策文本量化分析——基于PMC指数模型

新业态从业人员社会保障需充分且有效的政策体系支持,政策文本量化分析能够为未来政策制定与优化提供参考。基于政策制定前端视角,以中央和地方层面出台的41项新业态从业人员社会保障政策为研究对象,将政策文本挖掘和PMC指数模型相结合,建立新业态从业人员社会保障政策的量化评价指标体系,并从中选取有代表性的12项政策文本进行量化评价。研究发现:当前我国新业态从业人员社会保障政策文本处于“整体良好”水平;在评估的12项政策中,仅有1项优秀政策,其余7项良好政策和4项可接受政策均存在不同程度的优化空间;部分政策预测性和反馈机制不足,政策内容覆盖不全面,保障措施有待完善。据此,建议重视政策的前瞻性和反馈性,增加社会救助、部门协同和监督管理方面的内容,促进新业态从业人员社会保障政策精准发力。