CAJ | 학술논문

以微博用户推荐算法中相似度计算为研究对象，根据微博用户关注信息的特点，分析了关注用户的流行度的不同程度，以及这种程度差异对相似度计算产生的影响，在此基础之上提出了一种加入流行度制衡因子的相似度计算方法。可通过流行度制衡因子，在计算用户相似度时，适度减少（增加）流行度偏高（偏低）的用户对计算结果的影响。实验结果表明：加入流行度制衡因子的用户相似度计算具有更好的推荐效果。
이미박용호추천산법중상사도계산위연구대상，근거미박용호관주신식적특점，분석료관주용호적류행도적불동정도，이급저충정도차이대상사도계산산생적영향，재차기출지상제출료일충가입류행도제형인자적상사도계산방법。가통과류행도제형인자，재계산용호상사도시，괄도감소（증가）류행도편고（편저）적용호대계산결과적영향。실험결과표명：가입류행도제형인자적용호상사도계산구유경호적추천효과。
This paper aims to improve the performance of micro-blog users′recommendation algorithm whose key issue is user similarity calculation .This paper firstly introduces current several user similarity calculation methods and their shortcomings.And then, based on the characteristics of micro-blog users, we discuss their concerning users′popularity of different level , which aims to explore their effects on the user similarity calculation .And on this basis , an improved similarity calculating method of considering depression factor is proposed .In the user similarity calculation , popularity balance factor can reduce/increase the impact of calculation results caused by the user high/low popularity . The experimental results indicate that the proposed user similarity calculation method has better recommendation results .