本文選題：信息擴(kuò)散模型 + 異構(gòu)��； 參考：《南京大學(xué)》2017年碩士論文

【摘要】：近些年來,隨著信息技術(shù)的大力發(fā)展,個(gè)體傳播信息和獲取信息的能力大幅提升,隨之而來的是信息擴(kuò)散的影響和頻率也越來越大,所以對(duì)信息擴(kuò)散進(jìn)行分析研究以試圖控制信息擴(kuò)散過程具有重大的意義和價(jià)值�，F(xiàn)有的信息擴(kuò)散模型可以分為兩類:類傳染病模型和閾值模型。類傳染病模型是一類從經(jīng)典的SIR傳染病模型中類比拓展而來的信息擴(kuò)散模型,該類模型將傳播主體劃分成幾個(gè)狀態(tài),并定義狀態(tài)之間的轉(zhuǎn)化概率。閾值模型又被稱為基于決策的模型,在閾值模型中,傳播主體會(huì)依據(jù)其鄰居的決策進(jìn)行權(quán)衡比較,最終做出自己的決策。這些年來,在信息擴(kuò)散的研究中,這兩類模型都涌現(xiàn)了大量的、適用于各種各樣環(huán)境的研究成果。但是,這些已有的模型中存在以下兩點(diǎn)不足。第一,很少有研究者考慮到信息擴(kuò)散中傳播主體的多元性,特別是傳播主體在行為模式上的多元性。第二,現(xiàn)有的模型中有很多研究謠言傳播的模型,但幾乎沒有針對(duì)傳言的模型。傳言是一種非官方傳播的、未能辨別真?zhèn)蔚�、在現(xiàn)實(shí)中很常見的信息類別,而謠言是確定為假的傳言。針對(duì)上述問題,本文針對(duì)傳言,同時(shí)考慮了傳播主體在行為模式和自身屬性上的多元性,提出了一個(gè)能夠完全展現(xiàn)這兩類主體多元性的Agent模型。在行為模式的多元性方面,受到俗語"謠言止于智者"的啟發(fā),本文將信息擴(kuò)散中的傳播主體按照行為模式分為兩類:"智者"和"愚人"。其中,"智者"只有在認(rèn)為信息可信且對(duì)信息感興趣時(shí)才會(huì)傳播信息;"愚人"則會(huì)忽略信息的可信度,只要對(duì)信息感興趣就會(huì)傳播信息。此處傳播的信息就是傳言,不能確定真假,只能交由傳播主體自行判斷。在主體屬性的多元性方面,本文考慮了個(gè)體的名聲對(duì)信息可信度的影響:名聲好的個(gè)體更容易讓人相信其所傳播的信息;而名聲差的個(gè)體往往讓人懷疑他所傳播的信息。之后,基于能夠模擬異構(gòu)主體("智者"和"愚人")的Agent模型,本文提出了一個(gè)異構(gòu)性多Agent信息擴(kuò)散模型——S-F模型。接著,本文按照控制變量的方法設(shè)計(jì)了 6個(gè)實(shí)驗(yàn),以探索信息擴(kuò)散過程中存在的規(guī)律。
[Abstract]:In recent years, with the rapid development of information technology, the ability of individuals to disseminate and obtain information has been greatly improved, followed by the increasing impact and frequency of information diffusion. Therefore, it is of great significance and value to analyze and study information diffusion in order to control the process of information diffusion. The existing information diffusion models can be divided into two categories: infectious disease model and threshold model. Infectious disease model is a kind of information diffusion model which is extended by analogy from the classical SIR infectious disease model. This kind of model divides the propagator into several states and defines the transformation probability between the states. The threshold model is also called the decision-based model. In the threshold model, the propagator will make his own decision according to the decision of his neighbor. In recent years, in the research of information diffusion, there have been a large number of research results of these two models, which can be applied to various environments. However, there are two shortcomings in these existing models. First, few researchers consider the diversity of the subject of communication in information diffusion, especially the diversity of the behavior pattern of the subject. Second, there are many existing models to study rumor propagation, but there are few models for rumors. Rumors are an unofficial type of information that fails to identify authenticity and is common in reality, while rumors are confirmed as false. In order to solve the above problems, this paper proposes a Agent model which can fully show the diversity of these two types of agents by considering the multiplicity of the agents in the behavior pattern and their own attributes. In terms of the diversity of behavior patterns, inspired by the saying "rumor ends in the wise man", this paper divides the subject of information diffusion into two categories according to the behavior pattern: "wise man" and "fool". Among them, "wise" will only spread information when it is believed to be credible and interested in information; "fool" will ignore the credibility of information, as long as interested in information will spread information. The information transmitted here is rumors, can not determine the truth and falsehood, can only be judged by the subject itself. In terms of the diversity of subject attributes, this paper considers the influence of an individual's reputation on the credibility of information: a person with good reputation is more likely to believe the information he transmits, while an individual with a poor reputation often makes people doubt the information transmitted by him. Then, based on the Agent model which can simulate the heterogeneous agents ("wise man" and "fool"), this paper proposes a heterogeneous multi-Agent information diffusion model-S-F model. Then, six experiments are designed according to the method of controlling variables to explore the law of information diffusion.
【學(xué)位授予單位】：南京大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：G206;TP393.09
，

本文編號(hào)：1956057