Li-Xin Zhong

Alleviating bias leads to accurate and personalized recommendation

Recommendation bias towards objects has been found to have an impact on personalized recommendation, since objects present heterogeneous characteristics in some network-based recommender systems. In this article, based on a biased heat conduction recommendation algorithm (BHC) which considers the heterogeneity of the target objects, we propose a heterogeneous heat conduction algorithm (HHC), by further taking the heterogeneity of the source objects into account. Tested on three real datasets, the Netflix, RYM and MovieLens, the HHC algorithm is found to present better recommendation in both the accuracy and diversity than two benchmark algorithms, i.e., the original BHC and a hybrid algorithm of heat conduction and mass diffusion (HHM), while not requiring any other accessorial information or parameter. Moreover, the HHC algorithm also elevates the recommendation accuracy on cold objects, referring to the so-called cold-start problem. Eigenvalue analyses show that, the HHC algorithm effectively alleviates the recommendation bias towards objects with different level of popularity, which is beneficial to solving the accuracy-diversity dilemma.

Time scales of epidemic spread and risk perception on adaptive networks

Incorporating dynamic contact networks and delayed awareness into a contagion model with memory, we study the spreading patterns of infectious diseases in connected populations. It is found that the spread of an infectious disease is not only related to the past exposures of an individual to the infected but also the time scales of risk perception reflected in the social-network adaptation. The epidemic threshold pc is found to decrease with the rise of the time scale parameter s and the memory length T, which satisfies the equation . Both the lifetime of the epidemic and the topological property of the evolved network are considered. The standard deviation σd of the degree distribution increases with the rise of the absorbing time tc, a power-law relation σd=mtcγ is found.

Redundant correlation effect on personalized recommendation

Abstract The high-order redundant correlation effect is investigated for a hybrid algorithm of heat conduction and mass diffusion (HHM), through both heat conduction biased (HCB) and mass diffusion biased (MDB) correlation redundancy elimination processes. The HCB and MDB algorithms do not introduce any additional tunable parameters, but keep the simple character of the original HHM. Based on two empirical datasets, the Netflix and MovieLens , the HCB and MDB are found to show better recommendation accuracy for both the overall objects and the cold objects than the HHM algorithm. Our work suggests that properly eliminating the high-order redundant correlations can provide a simple and effective approach to accurate recommendation.

Geography and distance effect on financial dynamics in the Chinese stock market

Geography effect is investigated for the Chinese stock market including the Shanghai and Shenzhen stock markets, based on the daily data of individual stocks. The stocks in the Shanghai city and the Guangdong province are found to greatly contribute to the Shanghai and Shenzhen markets in the geographical sector, respectively. By investigating a geographical correlation on a geographical parameter, the stock location is found to have an impact on the financial dynamics, except for the financial crisis time of the Shenzhen market. Stock distance effect is further studied, with the probability of the short distance observed to be much greater than that of the long distance. The distance is found to only affect the stock correlation of the Shanghai stock market, but has no effect on the Shenzhen stock market.

A personalized recommendation algorithm via heterogeneous heat conduction

Heat conduction analogous process has ever been introduced into information filtering named standard heat conduction (SHC) method, resulting in a highly personalized but less accurate recommendation. In order to improve the recommendation accuracy, different algorithms have been proposed, with typical examples to be the highly accurate mass diffusion (MD) method, and a both highly accurate and highly diverse biased heat-conduction method (BHC). These previous algorithms have not considered the rating effect, where ratings essentially depict how users like objects. In this article, we propose a heterogeneous heat conduction method (HHC), by taking the ratings as the weight of heat conduction, which thus generates a heterogeneous heat diffusion pattern. Experimental results obtained from the Movie Lens dataset show that, the HHC greatly enhances the recommendation accuracy against the SHC, with the improvement percentage to be 46.32%, and also elevates the recommendation accuracy against the MD as well as the BHC. Moreover, the HHC simultaneously outperforms the MD, and even the BHC in recommendation diversity.

Self-reinforcing feedback loop in financial markets with coupling of market impact and momentum traders

Abstract By incorporating market impact and momentum traders into an agent-based model, we investigate the conditions for the occurrence of self-reinforcing feedback loops and the coevolutionary mechanism of prices and strategies. For low market impact, the price fluctuations are originally large. The existence of momentum traders has little impact on the change of price fluctuations but destroys the equilibrium between the trend-following and trend-rejecting strategies. The trend-following herd behaviors become dominant. A self-reinforcing feedback loop exists. For high market impact, the existence of momentum traders leads to an increase in price fluctuations. The trend-following strategies of rational individuals are suppressed while the trend-following strategies of momentum traders are promoted. The crowd–anticrowd behaviors become dominant. A negative feedback loop exists. A theoretical analysis indicates that, for low market impact, the majority effect is beneficial for the trend-followers to earn more, which in turn promotes the trend-following strategies. For high market impact, the minority effect causes the trend-followers to suffer great losses, which in turn suppresses the trend-following strategies.

A minority game with expected returns for modeling stock correlations

Financial systems are complex systems which have been widely studied in recent years. We here propose a model to study stock correlations in financial markets, in which an agents expected return for one stock is influenced by the historical return of the other stock. Each agent makes a decision based on his expected return with reference to information dissemination and the historical return of the stock. We find that the returns of the stocks are positively (negatively) correlated when agents expected returns for one stock are positively (negatively) correlated with the historical return of the other. We provide both numerical and analytical studies and give explanations to stock correlations for cases with agents having either homogeneous or heterogeneous expected returns. The result still holds when other factors such as holding decisions and external events are included which broadens the practicability of the model.

A generalized public goods game with coupling of individual ability and project benefit

Facing a heavy task, any single person can only make a limited contribution and team cooperation is needed. As one enjoys the benefit of the public goods, the potential benefits of the project are not always maximized and may be partly wasted. By incorporating individual ability and project benefit into the original public goods game, we study the coupling effect of the four parameters, the upper limit of individual contribution, the upper limit of individual benefit, the needed project cost and the upper limit of project benefit on the evolution of cooperation. Coevolving with the individual-level group size preferences, an increase in the upper limit of individual benefit promotes cooperation while an increase in the upper limit of individual contribution inhibits cooperation. The coupling of the upper limit of individual contribution and the needed project cost determines the critical point of the upper limit of project benefit, where the equilibrium frequency of cooperators reaches its highest level. Above the critical point, an increase in the upper limit of project benefit inhibits cooperation. The evolution of cooperation is closely related to the preferred group-size distribution. A functional relation between the frequency of cooperators and the dominant group size is found.