Yuehui Jin

ESIS: Emotion-based spreader–ignorant–stifler model for information diffusion

Abstract With the proliferation of online social networks, users are willing to post messages sharing their statuses. A piece of information can include not only substantial news but also emotional expression. As messages are re-posted among users, large cascades are created and information is spread with such emotional expression. We propose an emotion-based spreader–ignorant–stifler (ESIS) model to simulate the process of information diffusion. The proposed model categorizes information cascades into fine-grained classes, and the proportion of retweets among users for one emotion as weights on edges is introduced. We conduct experiments with artificial and real social networks. The experimental results indicate that the probability of information adoption is based both on the spreading probability and retweeting strength among users. We verify the proposed model and predict the cascade size with a real-world dataset. Compared to the latest related models, i.e., the standard SIS model and the information cascade models, the proposed ESIS model demonstrates 11.8% and 16.5% performance improvements, respectively.

An emotion-based independent cascade model for sentiment spreading

An emotion-based independent cascade model is proposed to analyze the process of sentiment spreading.User features, structural features and tweet features are introduced into the learning model for finding the sentiment changes of retweets.The historical information based transforming weights are proposed for the sentiment prediction of retweets. Online social networks (OSNs) provide a platform for users to publish messages, by which users express their emotions on events or products. The phenomenon that emotions are spread by retweeting messages is referred to as sentiment spreading. In this paper, an emotion-based independent cascade model is proposed to study the process of sentiment spreading. The proposed model divides the process of sentiment spreading into three steps. First, propagation probabilities are introduced to predict whether users retweet messages. Second, a learning model taking account of user features, structural features, and tweet features is applied to predict whether emotions are changed after retweeting. Third, the transforming weights are calculated to predict what the sentiments of the retweets transform to. The experimental results on Sina Weibo demonstrated that the proposed model could achieve 15.78% and 4.9% performance improvements compared with two baseline methods.

Handoff prediction for data caching in mobile Content Centric Network

As real-time mobile multimedia becomes more and more popular in modern life, Quality of Service (QoS) has drawn more and more attention. IEEE 802.11 has been widely deployed to provide broadband wireless network access due to its low cost, simplicity and high bandwidth capacity. However, its poor network availability and frequent network disconnection in a mobile scenario challenge real-time services. This paper presents a caching and pre-fetching mechanism based on the usage of Content Centric Network (CCN) in mobile environments. The idea is to anticipate disconnections and prefetch the content that a user is expected to access in the near future. Hence, when to trigger the caching and pre-fetching module is an essential part of the objective. This paper focuses on the design and implementation of a disconnection predictor algorithm. The method is to predict the time before a terminal disconnects from an Access Point (AP) by monitoring Received Signal Strength Indicator (RSSI). This paper evaluates and compares the used algorithms by experiments.

Joint optimization of performance and economics in inter-domain traffic engineering

A major functionality of Inter-domain Traffic Engineering is to optimally allocate resources to meet traffic demand. We implement this functionality by introducing economics into the problem of performance optimization, achieving maximum performance while pursuing minimum economic costs. In this paper we propose the DOPE model (Dual-objective Optimization on Performance and Economics) as the foundation of building an Inter-domain Traffic Engineering architecture which optimizes performance and reduces economic costs simultaneously. We utilize the well-known concept of Nash Bargaining to optimize the performance so as to make the solution Pareto-efficient and fair. In addition we introduce the Revenue Sharing Contract in order to make the ISPs collaborate voluntarily to minimize the social economic costs. To protect ISPs from the leakage of sensitive information, we take a Decomposition Method, which combines the Sub-gradient Method with Lagrangian Relaxation Algorithm, and separates the overall dual-objective optimization problem into partial sub-problems. These sub-problems can be solved independently by ISPs. Our proposed approach is evaluated in several experiments with simulated networks. The results show that our approach converges quickly and performs as well as the centralized solution with full knowledge of the networks. Besides, our approach achieves both optimized performance and reduced economic costs, which is significantly better than existing unilateral methods. Therefore, we believe that our approach is an effective solution to the problem of Inter-domain Traffic Engineering, and will be highly adopted by ISPs when compared with existing approaches.

Instability of TCP-RED in a tandem network

Due to the quality of some services, jitter should be minimized in the network. To avoid jitter, the authors adopt a discrete time feedback system model for a TCP (Transmission Control Protocol) network with RED(Random Early Detection) in a tandem network. On the basis of this model, jitter is known that it is caused by the instability of the network. In order to analyze the instability, the model while one system parameter gradually changes and other system parameter keep still is simulated. Then it is obtained that bifurcation behavior occurs, leading to chaotic behavior while the system parameter increases. Hence, the new router should be set accurately when it is connected to the existed network. Moreover, it is asserted that our work will have instructive impact on configuring the router.

Rumor spreading with nonlinear infectivities in weighted networks

In the standard rumor spreading model, each node is treated equally and each link between two nodes, has the same weight. But in the real world, each node represents individuals. That means we should treat them respectively. So we introduce two weights for one link in the different parts during the spreading process. One is when an ignorant meets a spreader while the other is when a spreader encounters another spreader. Both of weights stand for the interaction between two connected nodes. We also analytically simulate the process of spreading rumors both on the small-world network and the scale-free network. First, we imitate the time evolution of three groups, namely, the ignorant, the spreader and the stifler. Second, we show the effects of parameters in the model.