基于分布式符号数据的混合推荐算法研究

A Hybrid Recommendation Algorithm Based On Distributed Symbol Data

传统的基于点数据的混合推荐算法鉴于用户偏好信息难以统筹而不能够从用户的角度建立模型,推荐的质量与效率受到影响。为能够更加具体的表征用户模型,以分布式符号数据为基础,建立用户积极子模型和消极子模型,通过分布式符号数据的距离度量计算出用户间的积极相似度和消极相似度,最后采用协同过滤算法为目标项目进行评分预测。将这种全新的混合推荐算法与传统推荐算法进行对比实验,结果表明,在一定的实验条件下,基于分布式符号数据的混合推荐算法优于传统的推荐算法。

It is difficult for the traditional hybrid recommendation algorithm based on point data to coordinate user preference information,therefor,a model cannot be built up from the user’s perspective,and the quality and efficiency of recommendation are affected.In order to represent the user model in a more specific way,positive and negative sub-models of users were established based on distributed symbol data.Then,the distance measurement of distributed symbol data was used to calculate the positive similarity and negative similarity between users.Finally,the collaborative filtering algorithm was used to predict the score of the target project.The results show that the hybrid recommendation algorithm based on distributed symbol data is superior to the traditional recommendation algorithm under specific experimental conditions.