基于离散选择模型的推荐系统改进算法

准确地预估用户的点击率,并根据该概率对商品排序以供用户选择在推荐系统领域有着重要的意义。推荐系统中常用的因子分解机等机器学习模型一般只考虑用户选择单个商品的概率,忽略了候选商品之间的相互影响,离散选择模型则考虑将商品候选集作为整体进行考虑。提出了使用深度学习模型来改进离散选择模型,模型使用相对特征层、注意力机制等网络结构帮助深度学习模型进行不同商品间的特征比较,研究结果表明引入离散选择模型的深度学习模型表现优于梯度提升决策树、因子分解机等模型。

An Improved Recommendation System Algorithm Based on Discrete Choice Model

It is of great significance in the field of recommendation system to accurately estimate the click-through rate of users and sort the products according to the probability for users to choose.Machine learning models such as factorization machine commonly used in recommendation systems generally only consider the probability that users choose a single product,ignoring the interaction between candidates,while discrete choice model consider the candidates as a whole.In this paper,a deep learning model is proposed to improve the discrete choice model.The model uses network structures such as relative feature layer and attention mechanism to help deep learning model to compare the features of different commodities.The results show that the performance of the deep learning model with discrete selection model is better than that of the gradient boosting decision tree and the factorization machine.