CAJ | 학술논문

针对传统的知识推荐算法存在用户冷启动和冷门物品推荐的问题，提出了一种基于三部图网络结构的知识推荐算法。在计算相似度时引入网络结构中的度，综合考虑项目的度和权值及标签的度和权值对推荐算法的影响。实验结果表明，该算法提高了推荐的个性化和多样性，有效地解决了用户冷启动和冷门物品推荐的问题，改善了推荐效果。
침대전통적지식추천산법존재용호랭계동화랭문물품추천적문제，제출료일충기우삼부도망락결구적지식추천산법。재계산상사도시인입망락결구중적도，종합고필항목적도화권치급표첨적도화권치대추천산법적영향。실험결과표명，해산법제고료추천적개성화화다양성，유효지해결료용호랭계동화랭문물품추천적문제，개선료추천효과。
This paper proposed a knowledge recommendation algorithm based on tripartite graphs network structure to solve the problem of the user cold start and less popular item recommendation.It embedded the degree of the network structure into the similarity degree,considered the project’degrees and weights and the label’degrees and weights impacting on the recommenda-tion algorithm.The experimental results show,the algorithm improves the personalized and diversity of recommendation,and solves the problem of user cold start and less popular item recommendation effectively,improves the recommendation effect.