本文選題：社交網(wǎng)絡(luò)　切入點(diǎn)：網(wǎng)絡(luò)輿論　出處：《北京交通大學(xué)》2016年博士論文　論文類型：學(xué)位論文

【摘要】：隨著互聯(lián)網(wǎng)在全球范圍內(nèi)的飛速發(fā)展,網(wǎng)絡(luò)媒體被公認(rèn)為是繼報(bào)紙、廣播、電視之后的“第四媒體”,網(wǎng)絡(luò)已成為反映社會(huì)輿情的主要載體之一。社交網(wǎng)絡(luò)是網(wǎng)民進(jìn)行信息交流的重要平臺(tái),其獨(dú)特的組織模式催生出網(wǎng)絡(luò)生態(tài)的復(fù)雜化和觀點(diǎn)傳播的日趨碎片化,是網(wǎng)絡(luò)輿論形成和傳播中不可忽視的力量。社交網(wǎng)絡(luò)具有便捷多樣的接入方法和更為靈活的交互參與方式,同時(shí)包含用戶個(gè)體的自主性、交互行為的復(fù)雜性和群體的異質(zhì)性等潛在特征。網(wǎng)絡(luò)輿論的形成、演化和傳播依賴于用戶之間的交互行為,然而傳統(tǒng)輿論研究難以真實(shí)反映社交網(wǎng)絡(luò)輿論中的用戶特性,由此,一系列針對(duì)網(wǎng)絡(luò)輿論中個(gè)體交互行為和觀點(diǎn)演化模式研究的新挑戰(zhàn)日漸凸顯：個(gè)體交互研究需要多角度、細(xì)粒度地刻畫(huà)其行為特性,例如用戶關(guān)系網(wǎng)絡(luò)結(jié)構(gòu)、交互決策的選擇偏好：觀點(diǎn)演化研究應(yīng)重視與用戶影響力和影響范圍的結(jié)合；輿論傳播演化研究應(yīng)體現(xiàn)自由觀點(diǎn)信息之間的分化競(jìng)爭(zhēng)等。鑒于此,本文結(jié)合交叉學(xué)科研究方法,圍繞社交網(wǎng)絡(luò)中個(gè)體交互行為,對(duì)社交網(wǎng)絡(luò)用戶數(shù)據(jù)采集與用戶特性實(shí)證分析、用戶交互行為建模、觀點(diǎn)交互機(jī)制與演化過(guò)程、競(jìng)爭(zhēng)信息傳播模式等問(wèn)題進(jìn)行了深入研究。論文采用實(shí)證數(shù)據(jù)分析、數(shù)學(xué)建模和計(jì)算機(jī)仿真相結(jié)合的研究方法,重點(diǎn)研究社交網(wǎng)絡(luò)中個(gè)體交互和觀點(diǎn)演化的微觀規(guī)則和宏觀規(guī)律,探索網(wǎng)絡(luò)輿論的形成機(jī)制,刻畫(huà)網(wǎng)絡(luò)輿情的演化過(guò)程。研究不僅能夠加深對(duì)復(fù)雜系統(tǒng)和網(wǎng)絡(luò)輿情本質(zhì)的認(rèn)識(shí),豐富社交網(wǎng)絡(luò)中個(gè)體交互行為、網(wǎng)絡(luò)觀點(diǎn)交互與演化和輿論形成與傳播等方面的理論,而且能夠在網(wǎng)絡(luò)輿論的及時(shí)應(yīng)對(duì)、合理引導(dǎo)等方面得到實(shí)際應(yīng)用。論文的研究工作得到了國(guó)家自然科學(xué)基金項(xiàng)目(No.61271308、No.61401015)、北京市重點(diǎn)實(shí)驗(yàn)室資助項(xiàng)目和北京市重點(diǎn)學(xué)科建設(shè)項(xiàng)目等項(xiàng)目的支持。論文的主要工作和創(chuàng)新點(diǎn)如下：1.單向關(guān)注關(guān)系是微博不同于其他社交網(wǎng)絡(luò)的關(guān)系結(jié)構(gòu)特征,改變了用戶獲取信息的方法和傳播信息的途徑,論文實(shí)證分析微博數(shù)據(jù),進(jìn)而發(fā)現(xiàn)單向關(guān)注對(duì)微博關(guān)系網(wǎng)絡(luò)結(jié)構(gòu)、用戶交互行為和個(gè)體偏好帶來(lái)影響。研究中采用基于網(wǎng)絡(luò)爬蟲(chóng)的數(shù)據(jù)采集框架獲取大量微博用戶數(shù)據(jù),實(shí)證分析發(fā)現(xiàn)：?jiǎn)蜗蜿P(guān)注是導(dǎo)致節(jié)點(diǎn)度分布冪律偏移的主要原因,雙向關(guān)注在用戶好友關(guān)系中所占比例較低,呈現(xiàn)負(fù)指數(shù)衰減特征;用戶交互行為存在明顯偏好,單向關(guān)注和雙向關(guān)注用戶轉(zhuǎn)發(fā)行為表現(xiàn)出隨用戶偏好冪律衰減特性；雙向關(guān)注關(guān)系盡管所占比例低,卻表現(xiàn)出高頻度的交互行為和優(yōu)先的用戶偏好,在用戶交互過(guò)程中影響顯著。研究加深對(duì)社交網(wǎng)絡(luò)中用戶關(guān)系和行為特性的認(rèn)識(shí),并為網(wǎng)絡(luò)輿論中的微觀個(gè)體研究提供理論依據(jù)和數(shù)據(jù)支撐。2.用戶交互行為在網(wǎng)絡(luò)輿論的形成過(guò)程中起重要作用,多方面研究已經(jīng)發(fā)現(xiàn)了個(gè)體復(fù)雜行為中的冪律特性和多模特性,然而卻難以采用數(shù)學(xué)方法系統(tǒng)地描述并解釋相關(guān)現(xiàn)象,論文建模微觀個(gè)體交互行為,描述并研究社交網(wǎng)絡(luò)用戶復(fù)雜交互行為及其時(shí)間特性。論文在實(shí)證分析用戶選擇偏好特性的基礎(chǔ)上,提出偏好優(yōu)先的興趣信息決策機(jī)制,建立基于選擇偏好的社交網(wǎng)絡(luò)用戶交互模型,研究分析觀察范圍和興趣概率在微觀個(gè)體交互作用中的具體表現(xiàn),最后通過(guò)微博用戶數(shù)據(jù)驗(yàn)證模型的有效性。研究發(fā)現(xiàn)交互環(huán)境的改變將直接導(dǎo)致不同模式的交互結(jié)果,固定觀察范圍對(duì)應(yīng)于交互行為的負(fù)指數(shù)特性,隨機(jī)觀察范圍則對(duì)應(yīng)于冪律特性。論文分析了興趣信息集合中偏好用戶的隨機(jī)化表現(xiàn),解釋了交互行為分布中平穩(wěn)期、衰減期和截?cái)嗥诘某梢?并給出了個(gè)體交互行為及其多模特征的數(shù)學(xué)描述方法,量化了在面對(duì)大量信息流時(shí)選擇偏好對(duì)交互行為的直接影響。論文研究能夠真實(shí)反映社交網(wǎng)絡(luò)中個(gè)體交互現(xiàn)象,深化了對(duì)微觀交互規(guī)律的認(rèn)識(shí),具有重要的理論和實(shí)際意義。3.傳統(tǒng)觀點(diǎn)交互研究局限于相鄰個(gè)體之間的簡(jiǎn)單交互規(guī)則,難以準(zhǔn)確描述社交網(wǎng)絡(luò)中輿論形成與演化過(guò)程,論文采用觀點(diǎn)動(dòng)力學(xué)建模方法,研究廣闊的信息獲取范圍和個(gè)體異質(zhì)性對(duì)輿論演化的影響。論文提出基于擴(kuò)展觀察范圍與社會(huì)影響力的觀點(diǎn)交互和演化模型,借鑒CODA思想描述用戶個(gè)體觀點(diǎn)；結(jié)合多數(shù)群體和少數(shù)群體的社會(huì)作用,提出同時(shí)考慮用戶觀點(diǎn)強(qiáng)度、觀點(diǎn)接觸直接性和作用用戶數(shù)量的群體觀點(diǎn)交互機(jī)制,擴(kuò)展了用戶個(gè)體的觀察范圍：并引入成功勸說(shuō)作用的影響力反饋累積機(jī)制,使得話題討論與觀點(diǎn)演化更加符合社交網(wǎng)絡(luò)中的實(shí)際特征。研究發(fā)現(xiàn)了個(gè)體交互強(qiáng)度和觀點(diǎn)穩(wěn)定性之間存在制衡關(guān)系,以及影響力逐漸累積表現(xiàn)出的異質(zhì)性特征。該方法建立起微觀個(gè)體交互和宏觀輿論演化的紐帶,結(jié)合用戶影響力和影響范圍,量化了突破局部鄰居的觀點(diǎn)交互作用,能夠?yàn)樯缃痪W(wǎng)絡(luò)中追蹤輿論動(dòng)態(tài)、分析網(wǎng)絡(luò)輿情等應(yīng)用領(lǐng)域提供理論支撐。4.信息之間的競(jìng)爭(zhēng)和碰撞是社交網(wǎng)絡(luò)中用戶自由表達(dá)觀點(diǎn)并廣泛參與信息傳播的結(jié)果,而目前國(guó)內(nèi)外對(duì)多元競(jìng)爭(zhēng)信息之間的交互作用和傳播模式鮮有研究,論文建立競(jìng)爭(zhēng)信息傳播模型,研究社交網(wǎng)絡(luò)中多元競(jìng)爭(zhēng)信息的交互作用和傳播演化過(guò)程。論文研究借鑒傳染病動(dòng)力學(xué)理論,抽象出網(wǎng)絡(luò)用戶面對(duì)競(jìng)爭(zhēng)信息時(shí)的四種狀態(tài),提出基于二元競(jìng)爭(zhēng)信息交互的SHIR傳播模型。研究中利用平均場(chǎng)理論,推導(dǎo)出刻畫(huà)節(jié)點(diǎn)傳播狀態(tài)隨時(shí)間演化的微分方程組。仿真分析表明：中立的猶豫狀態(tài)催化了話題在傳播過(guò)程中的充分討論,提高了整體信息的總覆蓋率；二元信息的競(jìng)爭(zhēng)作用表現(xiàn)為對(duì)內(nèi)部有限用戶資源的爭(zhēng)搶；優(yōu)勢(shì)信息主導(dǎo)了信息的傳播與演化過(guò)程,競(jìng)爭(zhēng)信息的終態(tài)信息密度比值與其穩(wěn)態(tài)轉(zhuǎn)移概率比值存在冪律關(guān)系。研究建立了競(jìng)爭(zhēng)信息交互與多元信息傳播之間的聯(lián)系,彌補(bǔ)了現(xiàn)有研究中信息交互與傳播分離的不足之處,為準(zhǔn)確研究分析多維度的復(fù)雜網(wǎng)絡(luò)輿情提供了切實(shí)可行的研究方法。
[Abstract]:With the rapid development of the Internet in the global scope, the network media is recognized as the newspapers, radio, television after the "fourth media", the network has become the main carrier of public opinion. One of the social network is an important platform for Internet users to exchange information, its unique organizational model spawned increasingly complex and fragmented the views of network ecology, is a force to be reckoned with the network of public opinion formation and dissemination. Social network has a variety of convenient access method is more flexible and interactive participation, also include autonomy of the individual users, such as the complexity and heterogeneity of the potential characteristics of group interaction behavior. The formation of the network of public opinion the evolution and propagation depends on the interactive behavior between users, but the traditional public opinion research cannot accurately reflect the characteristics of users, social network public opinion in the network and a series of needle The new challenge mode evolution of individual interactions views and actions become increasingly prominent: the individual interaction research needs multi angle, fine-grained description of its behavior, such as the user relationship network structure, the preference of interactive decision making: a study of the viewpoint of evolution should be combined with the user attention and influence the scope and impact of the evolution of public opinion should be reflected; between the view of freedom of information differentiation competition. In view of this, this paper combined with the interdisciplinary research methods, focuses on individual interactions in social networks, analysis of social network user data acquisition and user characteristics of empirical, user interaction modeling, view interactive mechanism and evolution process, in-depth study of competition mode of information communication and other issues. The the empirical analysis, research methods of mathematical modeling and computer simulation combined with the focus on social network interaction and individual view The micro and macro point rule evolution rule, explore the formation mechanism of the network public opinion, modeling the evolution of network public opinion. The research can not only deepen our understanding of the complex system and the essence of network public opinion, rich individual social network interaction behavior, network interaction and evolution of ideas and public opinion formation and dissemination of the theory, but also in a timely manner with the network of public opinion, the practical application of a reasonable guide and so on. The research work of this thesis is supported by the National Natural Science Fund Project (No.61271308, No.61401015), supported by the Key Laboratory of Beijing city and Beijing city of key discipline construction projects such as support. The main work and innovation are as follows: 1. the one-way attention relationship structure micro-blog features different from other social networks, has changed the way of user access to information and dissemination of information, the empirical analysis Micro-blog data, and then found unidirectional interest in micro-blog network structure, the impact of user interaction and individual preference. The acquisition of a large number of users of micro-blog data collection framework web crawler based on the empirical analysis found that the one-way attention is the main cause of the node degree distribution of power-law offset, two-way attention for users of friends in a low proportion, negative exponential decay characteristics; there are obvious preference of user interaction, one-way and two-way attention focused on user behavior with forwarding user preference power-law attenuation characteristics; double to concern the relationship despite low proportion, but showed a high frequency of interactions and the priority of the user preferences, significant effects in the user in the interactive process. The research to deepen understanding of the social network user relationship and behavior characteristics, and provide rationale for the study of the individuals in the Internet public opinion On the basis of theory and data support.2. user interactions play an important role in the formation process of the network of public opinion, many studies have found that the power-law characteristics and characteristics of individual behavior in the multimode complex, but it is difficult to use mathematical methods to describe and explain the related phenomenon, the modeling of individual interaction behavior, describe and study social complex network user interaction behaviors and time characteristics. In the empirical analysis based on the selection of user preference characteristics, proposed interest preference information in decision-making mechanism, establish social network user interaction model selection based on preference, research and analysis of specific performance observation range and interest probability in the micro individual interaction, finally through the effectiveness of micro-blog users to validate the model. The study found that interactive environment changes will lead directly to the interaction results of different models, fixed observation range The negative exponential characteristics corresponding to the interaction behavior of the random observation range corresponds to the power law. This paper analyzes the performance of user preference information in randomization in the collection, explains the interactive behavior of the stable distribution, attenuation and truncation cause period, and describe the method of individual interaction behaviors and characteristics of multimode mathematical quantification in the face of a lot of information flow preferences directly influence on the interactive behavior. This study can reflect the phenomenon of individual interactions in social networks, to deepen the understanding of the microscopic interaction rules, simple interaction rules has important theoretical and practical significance of.3. traditional view research was limited to the adjacent interaction between individuals, it is difficult to accurately describe the process with the evolution of public opinion formation in social networks, the opinion dynamics modeling method of wide scope of information acquisition and individual heterogeneity of public opinion The influence of evolution. In this paper extended view interaction and evolution model of the observation range with the influence of society based on the thought of using CODA to describe the individual user view; combined with the social role the majority and minority groups, put forward considering the user views the strength, viewpoint of population interaction mechanism and the role of direct contact point of the number of users, the expansion of the individual users the observation range and the introduction of successful persuasion: the influence of cumulative feedback mechanism, makes the topic of discussion and opinion evolution accords with the actual characteristics of social network. The study found there was a relationship between the individual interaction strength and point stability, heterogeneity and cumulative influence gradually shows. The method to establish ties and individual interaction macro public opinion evolution, combined with the user influence and scope, to quantify the local neighborhood point to break through The interaction of public opinion to track dynamic social network analysis, competition and collision between the.4. information theory is a social network user freedom of expression and broad participation in the information dissemination network public opinion and other applications, while at home and abroad to study the interaction and communication between multiple few competitive information, the establishment of the competition information dissemination model, interaction and dissemination of information on diverse competition in social network evolution. Infectious disease dynamics theory research, abstract network users face four kinds of competition information, put forward the SHIR propagation model two yuan competition based on information interaction. Based on mean field theory, differential equations describes propagation state with time evolution is derived. The simulation analysis shows that the neutral state of hesitation in the dissemination process of the topic of catalysis The full discussion, to improve the overall information of the total coverage of two yuan; competition information for internal users for information resources; advantages dominated the propagation and evolution process of information, competition information terminal state information and the steady transition probability density ratio the ratio of the power law relation is established between the competitive information research. Interact with multiple communication links, make up for the deficiencies of the separation of information exchange and spreading in the present research, the research provides a feasible method for accurate analysis of the complex network of public opinion in various dimensions.

【學(xué)位授予單位】：北京交通大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2016
【分類號(hào)】：TP393.09
，

本文編號(hào)：1634791