Memory effect of the online user preference

Abstract

The mechanism of the online user preference evolution is of great significance for understanding the online user behaviors and improving the quality of online services. Since users are allowed to rate on objects in many online systems, ratings can well reflect the users' preference. With two benchmark datasets from online systems, we uncover the memory effect in users' selecting behavior which is the sequence of qualities of selected objects and the rating behavior which is the sequence of ratings delivered by each user. Furthermore, the memory duration is presented to describe the length of a memory, which exhibits the power-law distribution, i.e., the probability of the occurring of long-duration memory is much higher than that of the random case which follows the exponential distribution. We present a preference model in which a Markovian process is utilized to describe the users' selecting behavior, and the rating behavior depends on the selecting behavior. With only one parameter for each of the user's selecting and rating behavior, the preference model could regenerate any duration distribution ranging from the power-law form (strong memory) to the exponential form (weak memory).

Evolution properties of the community members for dynamic networks

“…In addition, tracing the community evolution is a challenge work, using the Markov process to describe the evolution of community structure [29] is an important method for this problem. Then, during the community evolution, how to explore the importance of nodes for dynamic networks [30][31][32][33] is also important problem to understand deeply the structure of networks.…”

Evolution properties of online user preference diversity

“…Processing the user rating sequence is reasonably regarded as the direct and effective way to reflect the properties of online user preferences. For instance, consecutively delivering higher or lower ratings has been used to identify the anchoring bias and memory effect of user online rating and selecting behaviors [24,25]. In this paper, the evolution properties of online user preference diversity could be measured by the variation coefficient V , a widely used distribution-based measure in the field of complex systems [37], defined as the ratio of the standard deviation σ to the mean μ.…”

“…These data sets are widely applied to modeling the patterns of online user preferences [24][25][26]. Each record in these data sets reads the user-object pair, followed by the rating the user gave to the object as well as the corresponding timestamp.…”

“…However, quite a lot of recent advances demonstrated that the user rating and selecting behaviors present considerable correlations with the historical ones (e.g., Anchoring bias effect and Memory effect of online user rating behaviors), indicating the Markovian property implicit in the user rating sequence [24,25,29,30]. Therefore, in this paper, we try to regenerate the dynamical rating process for each data set with a Markov model, where the relationship between the user's two consecutive actions are calculated based on the transition matrix, in which the distribution of the transition probability is captured by the Gaussian distribution [25,26]. Suppose the user u delivered m u (t) ratings during a given time interval […”

“…Yang et al found that there is an anchoring bias effect during the user rating process [24]. Hou et al [25] investigated the memory effect of the online user rating series [26,27] and found that the Markov model could regenerate online user collective behaviors, which indicates that a user's next action depends only on his/her current behaviors [28][29][30][31].…”