计算机应用与软件
計算機應用與軟件
계산궤응용여연건
COMPUTER APPLICATIONS AND SOFTWARE
2015年
6期
83-86
,共4页
LBS%社会网络%标签%个性化推荐
LBS%社會網絡%標籤%箇性化推薦
LBS%사회망락%표첨%개성화추천
Location-based service (LBS)%Social network%Tag%Personalised recommendation
针对现有的基于 LBS(Location Based Service)个性化推荐系统在构建用户兴趣模型时存在的缺陷,提出一种综合 LBS 和社会网络标签的个性化推荐(LTCF)方法。通过引入网络标签和用户社会关系,从用户标注的标签资源中找到拥有共同兴趣爱好的用户关系以及从社会网络中找到与目标用户关系紧密的用户,同时结合考虑用户兴趣爱好随空间不断变化的特点,依据协同过滤算法,计算用户社会关系度和用户空间相似性,依此得到目标用户的最近邻集合,在最近邻集基础上给出推荐结果。实验结果表明,相比于传统的基于 LBS 推荐方法,LTCF 模型在查全率和产准率有了显著的提升,能更好地反映出用户偏好,显著提高了推荐准确度。
針對現有的基于 LBS(Location Based Service)箇性化推薦繫統在構建用戶興趣模型時存在的缺陷,提齣一種綜閤 LBS 和社會網絡標籤的箇性化推薦(LTCF)方法。通過引入網絡標籤和用戶社會關繫,從用戶標註的標籤資源中找到擁有共同興趣愛好的用戶關繫以及從社會網絡中找到與目標用戶關繫緊密的用戶,同時結閤攷慮用戶興趣愛好隨空間不斷變化的特點,依據協同過濾算法,計算用戶社會關繫度和用戶空間相似性,依此得到目標用戶的最近鄰集閤,在最近鄰集基礎上給齣推薦結果。實驗結果錶明,相比于傳統的基于 LBS 推薦方法,LTCF 模型在查全率和產準率有瞭顯著的提升,能更好地反映齣用戶偏好,顯著提高瞭推薦準確度。
침대현유적기우 LBS(Location Based Service)개성화추천계통재구건용호흥취모형시존재적결함,제출일충종합 LBS 화사회망락표첨적개성화추천(LTCF)방법。통과인입망락표첨화용호사회관계,종용호표주적표첨자원중조도옹유공동흥취애호적용호관계이급종사회망락중조도여목표용호관계긴밀적용호,동시결합고필용호흥취애호수공간불단변화적특점,의거협동과려산법,계산용호사회관계도화용호공간상사성,의차득도목표용호적최근린집합,재최근린집기출상급출추천결과。실험결과표명,상비우전통적기우 LBS 추천방법,LTCF 모형재사전솔화산준솔유료현저적제승,능경호지반영출용호편호,현저제고료추천준학도。
In light of the shortcomings of existing LBS-based personalised recommendation system in building user interests model,this pa-per proposes a personalised recommendation method which integrates the LBS and social network tag (LTCF).By introducing network tag and users social relations,the method finds from the tag resources marked by users the relationship between users having common interests and hobbies as well as finds from social networks the users having intimate relationship with targeted users.Meanwhile,combining the considera-tion of the feature that users'interests constantly change along with the space and in accordance with collaborative filtering algorithm,the method calculates the social relation degree of users and the similarity of user spaces,then according to these it obtains the nearest neighbour sets of target users,and provides the recommendation results based on the nearest neighbour sets.Experimental results show that compared with traditional LBS-based recommendation method,the LTCF model has noticeable improvement in recall rate and precision product rate and can better reflect users'preferences as well as significantly improve the recommendation accuracy.