基于贝叶斯动态预测模型的商品推荐方法

An Approach to Commodity Recommendation Based on Bayesian Dynamic Forecasting Model

传统的电子商务推荐系统虽然考虑到个性化的推荐,但不能很好的描述用户行为,使得个性化的推荐略显不足。本文提出基于贝叶斯动态预测的模型,并结合Agent技术,很好地建立了用户行为预测模型。该方法以用户历史数据为基础,并结合用户的实时行为建立用户行为预测模型。本文将此方法运用于商品推荐系统中,实验证明此方法能高效地为客户产生个性化的商品推荐集合,优于某些传统方法。

Although typical recommendation on E-commerce takes individual recommendation into consideration, it cannot describe users’ behavior very well so as to make individual recommendation to run poorly. This paper build a users’ behavior model based on Bayesian dynamic forecasting model with agent techniques, the model is built by learning from users’ history data and behaviors at present. This method is used in a commodity recommendation system, an experimental result demonstrates that this method can effectively generate an individual recommendation set of commodity, and it is better than some traditional methods.