On the Volatility of Online Ratings: An Empirical Study

Online Reputation Model Using Moving Window

“…Naïve averaging method is the common method used for aggregating ratings on electronic social and commerce systems [10]. This method is not informative as it cannot discover recent trend in ratings and easily influenced by unfair ratings [10].…”

“…This method is not informative as it cannot discover recent trend in ratings and easily influenced by unfair ratings [10]. In turns, the weighted average method seems more effective as it can take into consideration the quality of users who made ratings.…”

“…On the other hand, another study suggested using the number of past transaction instead of ratings age [12]. Leberknight et al [10] reveled that recent ratings should have higher weight than old ratings, and the reputation model should take that as discounting factor during ratings aggregation process. They proposed a model that divides rating into number of non-overlapping equal subsets, and then investigate the volatility in each subset with respect to the near subset.…”

“…E-commerce and mobile commerce systems are increasingly growing in the last two decades which resulted in emergence of new technologies and services [1][6] [10]. Therefore, the internet turns into the most common workspace for performing our transactions such as selling and purchasing goods.…”

“…Many online rating systems use today the simplest method for aggregating rating which is the naïve average method. This method is straightforward, but it cannot discover the trends emerging from recent ratings [10]. To illustrate that, let us have a look at Figure 1 which illustrates that there is a clear trend in ratings where recent ratings are lower than old ratings.…”