本文關(guān)鍵詞： 大數(shù)據(jù) 網(wǎng)絡(luò)輿情 研判模型 應(yīng)對(duì)策略　出處：《東南大學(xué)》2016年博士論文　論文類(lèi)型：學(xué)位論文

【摘要】：隨著各種智能移動(dòng)終端設(shè)備的普及以及各種即時(shí)通訊軟件和平臺(tái)的發(fā)展,我國(guó)互聯(lián)網(wǎng)進(jìn)入Web2.0時(shí)代�；ヂ�(lián)網(wǎng)改變了傳統(tǒng)的輿情表現(xiàn)方式,把網(wǎng)絡(luò)輿情推到了反映民眾情緒和行為傾向的前臺(tái)。網(wǎng)絡(luò)平臺(tái)以開(kāi)放的空間形態(tài),成為個(gè)體和社會(huì)組織參政議政、表達(dá)態(tài)度、發(fā)表言論的公共平臺(tái),成為快速傳遞信息和傳達(dá)民意的通道,成為各種社會(huì)思潮交鋒、各種利益訴求集散和多種意識(shí)形態(tài)較量的陣地,網(wǎng)絡(luò)輿情研判和應(yīng)對(duì)成為當(dāng)今網(wǎng)絡(luò)社會(huì)一項(xiàng)新的重要任務(wù)。當(dāng)前網(wǎng)絡(luò)輿情研究,面臨的主要問(wèn)題是信息冗余和信息傳播方式革命性變化所帶來(lái)的夾雜大量噪音的海量數(shù)據(jù)的處理,導(dǎo)致了基于傳統(tǒng)的數(shù)據(jù)挖掘技術(shù)無(wú)法適應(yīng)新的要求,而且缺乏對(duì)不同網(wǎng)絡(luò)輿情的細(xì)分,缺乏針對(duì)不同特質(zhì)的網(wǎng)絡(luò)輿情建立不同的分析模型進(jìn)行分析,目前市場(chǎng)上的網(wǎng)絡(luò)輿情分析軟件以同一模型籠統(tǒng)應(yīng)對(duì)不同特征的網(wǎng)絡(luò)輿情,存在較大局限性。本文以大數(shù)據(jù)環(huán)境為背景,在對(duì)國(guó)內(nèi)外相關(guān)研究現(xiàn)狀進(jìn)行歸納分析的基礎(chǔ)上,主要針對(duì)網(wǎng)絡(luò)謠言、高校學(xué)生網(wǎng)絡(luò)輿情和突發(fā)公共衛(wèi)生事件等三類(lèi)典型的網(wǎng)絡(luò)輿情,采用定性分析和定量分析相結(jié)合的方法,圍繞網(wǎng)絡(luò)輿情的傳播機(jī)制、預(yù)警決策機(jī)制和演化機(jī)理開(kāi)展一系列的研究。在基于模型分析的基礎(chǔ)上,提出針對(duì)不同類(lèi)型網(wǎng)絡(luò)輿情的管理和應(yīng)對(duì)策略。首先,基于傳染病動(dòng)力學(xué)理論,文中構(gòu)建了具有飽和接觸率的網(wǎng)絡(luò)謠言傳播研判模型和非線性接觸率網(wǎng)絡(luò)謠言傳播研判模型,利用動(dòng)力系統(tǒng)平衡點(diǎn)理論與穩(wěn)定性理論,對(duì)網(wǎng)絡(luò)謠言進(jìn)行了定量分析。研究結(jié)果表明,在網(wǎng)絡(luò)謠言傳播中存在一個(gè)閾值R0,當(dāng)R01時(shí),系統(tǒng)將存在內(nèi)部非零平衡點(diǎn),即如任由謠言發(fā)展,會(huì)在系統(tǒng)中大面積爆發(fā)開(kāi)來(lái);當(dāng)網(wǎng)民群體人數(shù)服從Logistic曲線時(shí),新增加的網(wǎng)民不會(huì)對(duì)網(wǎng)絡(luò)謠言的傳播造成影響;由于閾值對(duì)心理作用系數(shù)的變化非常敏感,因此采取措施增大心理作用系數(shù)可以高效管理網(wǎng)絡(luò)謠言的擴(kuò)散;披露不實(shí)信息以及不實(shí)信息傳播者,其管理效率要遠(yuǎn)高于正向宣傳。其次,針對(duì)突發(fā)公共衛(wèi)生事件網(wǎng)絡(luò)輿情傳播的特點(diǎn),引入Deffault模型,建立了有向加權(quán)動(dòng)態(tài)網(wǎng)絡(luò)結(jié)構(gòu)模型,利用Matlab工具對(duì)所建立的網(wǎng)絡(luò)輿情觀點(diǎn)演化模型進(jìn)行仿真分析,驗(yàn)證了所建立的模型的有效性和合理性,還研究了影響網(wǎng)絡(luò)輿情觀點(diǎn)演化傳播的主要因素。結(jié)果表明,有向加權(quán)動(dòng)態(tài)BBV網(wǎng)絡(luò)是無(wú)標(biāo)度網(wǎng)絡(luò),符合在線社會(huì)網(wǎng)絡(luò)結(jié)構(gòu)的特性。模型分析還發(fā)現(xiàn),政府的態(tài)度r、媒體的關(guān)注程度λ等都能對(duì)網(wǎng)絡(luò)輿情產(chǎn)生顯著影響。因此,政府及主要公眾媒體利用自身權(quán)威性及時(shí)披露信息,加強(qiáng)疏導(dǎo),可以有效消除社會(huì)恐慌,穩(wěn)定社會(huì)局面。接著,針對(duì)高校學(xué)生網(wǎng)絡(luò)輿情預(yù)警級(jí)別的評(píng)判,構(gòu)建了基于直覺(jué)模糊推理和層次分析法的網(wǎng)絡(luò)輿情定性和定量評(píng)判模型。關(guān)于運(yùn)用直覺(jué)模糊推理判斷網(wǎng)絡(luò)輿情預(yù)警等級(jí),將話題重要性、公眾反應(yīng)和公眾與話題聯(lián)系作為直覺(jué)模糊推理的參與因素,用直覺(jué)模糊綜合評(píng)判法計(jì)算每個(gè)因素的隸屬度,將最貼近的直覺(jué)模糊集作為網(wǎng)絡(luò)輿情預(yù)警等級(jí),運(yùn)用直覺(jué)模糊集理論構(gòu)建了網(wǎng)絡(luò)輿情預(yù)警級(jí)別判定模型。對(duì)于運(yùn)用層次分析法判定網(wǎng)絡(luò)輿情預(yù)警等級(jí),利用層次分析法將目標(biāo)分解為多指標(biāo)層次,引入專(zhuān)家打分法確定各級(jí)指標(biāo)權(quán)重,構(gòu)造了反映高校網(wǎng)絡(luò)輿情傳播深度和廣度的定性與定量相結(jié)合的指標(biāo)體系,在對(duì)各級(jí)指標(biāo)具體權(quán)重值進(jìn)行一致性檢驗(yàn)后,根據(jù)所構(gòu)建的模型計(jì)算網(wǎng)絡(luò)輿情研判的指標(biāo)值S,根據(jù)S值所對(duì)應(yīng)的閾值區(qū)間,確定應(yīng)啟動(dòng)的預(yù)警級(jí)別,進(jìn)而通過(guò)分析其變化的基本特征,掌握其發(fā)展態(tài)勢(shì),揭示出問(wèn)題的本質(zhì)所在,預(yù)測(cè)出輿情的進(jìn)一步走向,可以幫助決策者做出正確決策,對(duì)輿論進(jìn)行引導(dǎo)和控制。實(shí)證研究表明,以網(wǎng)絡(luò)數(shù)據(jù)的收集整理和專(zhuān)家決策人員的理性判斷為切入點(diǎn),通過(guò)定量和定性相結(jié)合,可以及時(shí)準(zhǔn)確地判斷輿情級(jí)別,為及早啟動(dòng)預(yù)警流程和進(jìn)行引導(dǎo)干預(yù),有效控制輿情發(fā)展態(tài)勢(shì)提供支持。本文最后還進(jìn)行了案例分析。選取天津?yàn)I海新區(qū)爆炸事件、湖南大學(xué)研究生違規(guī)轉(zhuǎn)學(xué)事件作為典型案例,以本文中的理論研究為基礎(chǔ),研究了網(wǎng)絡(luò)謠言的傳播機(jī)制、網(wǎng)絡(luò)輿情的預(yù)警機(jī)制和網(wǎng)絡(luò)輿情意見(jiàn)的演化過(guò)程,分別采集謠言和公共衛(wèi)生影響的關(guān)鍵詞,對(duì)事件進(jìn)行描述,將特征數(shù)據(jù)代入模型進(jìn)行求解,并對(duì)結(jié)果進(jìn)行分析。研究結(jié)果表明,不同類(lèi)型的網(wǎng)絡(luò)事件具有較為明顯的內(nèi)在規(guī)律和特點(diǎn),本文所建立網(wǎng)絡(luò)輿情研判模型是有效的。
[Abstract]:With a variety of intelligent mobile terminal equipment and a variety of popular instant messaging software platform and the development of China's Internet into the Web2.0 era. The Internet has changed the traditional public opinion expression, network public opinion to reflect public sentiment and behavior tendency of the front desk. The network platform to open space, become the individual and social organization participation besides, the expression of attitude, the public platform of speech, become the rapid transmission of information and communication channels of public opinion, as against a variety of social thought, various interests distribution and various ideological battle positions, network public opinion judged and respond to today's social network become a new important task. The current network public opinion research, is the main problem facing with a lot of noise processing of massive data of information redundancy and information dissemination way brings revolutionary change, based on the traditional lead The data mining technology can not adapt to the new requirements, and the lack of network public opinion segmentation, lack of public opinion against the network of different characteristics of different models of analysis, the current network of public opinion on the market analysis of network public opinion in the same general software model with different characteristics, there is a big limitation. Based on the background of large data environment, the foundation of analysis on the related research at home and abroad, mainly for Internet rumors, the college students network public opinion and public health emergencies and other three kinds of typical network public opinion, using the method of qualitative analysis and quantitative analysis, on the transmission mechanism of network public opinion, to carry out a series of studies on the early warning decision the mechanism and evolution mechanism. Based on the model analysis, proposed management and coping strategies of different types of network public opinion. Firstly, based on the Epidemic dynamics theory, this paper constructs a network spread rumors saturated contact rate evaluation model and nonlinear contact rate of network rumor judged by dynamic system model, equilibrium theory and stability theory, the network rumors are quantitatively analyzed. The results show that there is a threshold of R0 in the network spread rumors, when R01 when the system will exist within the nonzero equilibrium points, such as let the rumor in the system development, will the outbreak of a large area of it; when users group number obeys the Logistic curve, the impact of the increase in the spread of new users will not be on the network rumors; because the threshold is very sensitive to changes on the psychological factor, so take measures to increase the psychological effect of diffusion coefficient can be the efficient management of network rumor; the disclosure of false information and false information spread, the management efficiency is much higher than that of the positive publicity. Secondly, according to the The characteristics of public health emergency network public opinion dissemination, the introduction of the Deffault model, based on weighted dynamic network structure model, on the view of network public opinion evolution model is analyzed and simulated by using Matlab tools, the results prove that the model is effective and reasonable, also studied the main factors affecting the evolution of the network public opinion dissemination view. The results show that the directed weighted dynamic BBV network is a scale-free network, with the characteristics of online social network structure model. The analysis found that the attitude of the government of R, the media attention degree lambda can have a significant impact on the Internet public opinion. Therefore, the government and public media using its own authority timely disclosure of information and strengthen counseling, can effectively eliminate social panic, stable social situation. Then, according to the university student network public opinion warning level evaluation, construct intuitionistic fuzzy reasoning based on The network public opinion qualitative and quantitative evaluation model analysis method and hierarchy. On the application of intuitionistic fuzzy reasoning to judge the warning level of the network of public opinion, the topic significance, the reaction of the public and public topics and links as intuitionistic fuzzy reasoning participation factor, membership in intuitionistic fuzzy comprehensive evaluation method to calculate each factor, will be the most close to the intuitionistic fuzzy sets as the warning level of network public opinion, fuzzy set theory to construct the warning level network public opinion judgment model for use of intuition. Determined by using AHP and warning level network public opinion, using AHP method to decompose the goal for the multi index level, introducing expert scoring method to determine the weights at all levels, construct the index system to reflect the qualitative and quantitative analysis of university network public opinion dissemination the depth and breadth of the combination, in the specific weight of all index value of the consistency test, according to the construction of Calculation model of network public opinion judged the index value of S, according to the S value corresponding to the threshold interval should be determined to start early warning level, and then through the analysis of the basic characteristics of the changes in the grasp of its development trend, reveal the essence of the problems and predict further towards the public opinion, can help decision-makers to make correct decisions on public opinion to guide and control. The empirical study shows that in a rational judgment of network data collection and expert decision makers as the starting point, through a combination of quantitative and qualitative, can timely and accurately determine the level of public opinion, to start early warning process and guide intervention, effective control of support the development trend of public opinion. In the end of case analysis. Taking Tianjin Binhai New Area bombing, graduate students of Hunan University as a typical case of illegal transfer events, based on theoretical research in this paper is based on The network rumor spreading mechanism, the evolution process of the early warning mechanism of network public opinion and network public opinion, keyword rumors and public health impact were collected. The description of the event, will feature data into the model, and the results were analyzed. The results show that different types of network events have the intrinsic rules and characteristics obviously, this network of public opinion judged model is effective.

【學(xué)位授予單位】：東南大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2016
【分類(lèi)號(hào)】：D669;C912.63
，

本文編號(hào)：1454352