基于事件本体的商品个性化推荐算法仿真

Simulation of Personalized Recommendation Algorithm Based on Event Ontology

针对当前电商网络商品推荐准确率低的问题,研究了基于事件本体的商品个性化推荐算法,以期提升商品推荐排名以及推荐丰富度。首先分析事件本体中上层的事件类关系结构和下层的事件推理与通用分类结构,并构建事件本体商品标签模型,规范化处理并分类商品标签。然后构建用户偏好模型,计算用户使用标签的频率权重衡量用户偏好,通过计算用户偏好与事件本体商品标签之间的关联程度分类用户与商品标签。将计算得到相似度较高的商品标签标注的商品,经协同过滤推荐算法推荐给用户。经验证可知,上述算法的推荐准确性随着TOP推荐的数量增加而增大,且所推荐的商品排序都最靠前,推荐商品内容最丰富,可以满足用户个性化多元需求。

Due to the problem of low accuracy of commodity recommendation in the current e-commerce network,this paper proposed an algorithm of personalized recommendation based on event ontology in order to improve the commodity recommendation ranking and make it more and more abundant.Firstly,the relation structure of the upperlevel event class and the lower-level event reasoning and general classification structure was analyzed,and then a model of product label of event ontology was constructed to process and classify these labels in a normalized way.After that,a user preference model was built to calculate the weight of the users frequency of using a label and thus to measure the users preference.After calculating the correlation degree between the users preference and the product label of event ontology,users and product labels could be classified.Based on the calculation result,the products with high similarity tags were recommended to users by the collaborative filtering recommendation algorithm.The simulation result shows that the recommendation accuracy of the proposed algorithm increases with the increase of the number of TOP recommendations,and the recommended products are very highly ranked.In addition,these products have substantial content,which can meet userspersonalized and diverse needs.