摘要
[目的/意义]社会化标注的标签中"涌现"的语义反映的是参与协作标注活动的社群普遍共识的知识,可以用来增强情境敏感型群偏好预测的准确性,但没有参与标注的用户,可能并不认同这种共识,基于社群涌现语义推理该用户偏好就会有偏差。因此研究针对用户的涌现语义适用性和其在情境敏感型群体偏好预测中的应用,以降低偏好预测的可能偏差,提升面向群体的推荐服务的质量。[方法/过程]以群体为信息服务对象,将基于个体实际评价反馈的情境相似度,和基于涌现语义的情境相似度进行相关性分析,度量涌现语义对当前用户的适用性,然后提出群偏好预测中涌现语义的整合应用模式。[结果/结论]实验结果表明,所提的涌现语义适用度应用模式,能提升情境敏感型群偏好预测的准确率,有利于面向群体的资源推荐精准性。[局限]实验数据不够丰富多样,存在稀疏性,涌现语义适用度阈值的调校方法和运算效率问题尚没有进行研究。
[Purpose/significance] The emergent semantic from social tags is the knowledge representing general consensus of the community involved in collaborative labeling activities,which can be used for the improvement of context-aware group preference prediction. But users who are not involved in the labeling activity may not agree with this consensus,therefore user preference based on community emergent semantic reasoning may be biased. In view of this problem,the paper tries to reduce the bias and improve the quality of recommendation service for community groups. [Method/process] Taking groups as information service objects,the paper makes a correlation analysis of context similarity between users' actual evaluation feedback and emergent semantics,measures the semantic applicability for current users,and proposes the integrated application model of the emergent semantic in group preference prediction. [Result/conclusion]The experiment results show that the proposed emergent semantic applicability model can improve the prediction accuracy of context-aware group preference,which is good for the accuracy of resources recommendation for the public. [Limitations]The experiment dataset is sparseness and not rich and diversified enough. Also,the adjustment methods of emergent semantic applicability's threshold and the efficiency of the algorithm have not been studied.
引文
[1]KIM J K,KIM H K,OH H Y,et al.A group recommendation system for online communities[J].International Journal of Information Management,2010,30(3):212-219.
[2]胡慕海,蔡淑琴,张宇,等.情境化信息推荐机制的研究[J].情报学报,2011,10(10):1053-1064.
[3]唐晓波,全莉莉.基于分众分类的本体构建分析[J].情报理论与实践,2008,31(6):133-138.
[4]ABERER K,CUDRMAUROUX P,OUKSEL A M,et al.Emergent semantics principles and issues[J].Database Systems for Advanced Applications,2004,2973:25-38.
[5]KOMPATSIARIS I,DIPLARIS S,PAPADOPOULOS S.Extracting emergent semantics from large-scale user-generated content[J].Advances in Intelligent&Soft Computing,2012,150:27-37.
[6]刘凯鹏,方滨兴.基于社会性标注的本体学习方法[J].计算机学报,2010,33(10):1823-1834.
[7]张云中,李佳佳.国外Folksonomy与Ontology融合研究的热点与趋势[J].图书馆理论与实践,2016(7):39-44.
[8]SUN L,WANG X,WANG Z,et al.Social-aware video recommendation for online social groups[J].IEEE Transactions on Multimedia,2016,19(3):609-618.
[9]ZOMAHOUN D E,YETONGNON K.EMERGSEM:emergent semantic and recommendation system for image retrieval[C]//Tenth International Conference on Signal-Image Technology and Internet-Based Systems.IEEE Computer Society,2014:256-263.
[10]丛维强,李保珍,王逊.基于加权元组潜在语义分析的社会标签推荐[J].情报科学,2015(1):57-62.
[11]EFTHYMIOU V,ZERVOUDAKIS P,STEFANIDIS K,et al.Group recommendations in mapreduce[J].2017,18(3):210-217.
[12]NAJJAR N A,WILSON D C.Evaluating group recommendation strategies in memory-based collaborative filtering[C]//ACMRecommender Systems Conference Workshop on Human Decision Making in Recommender Systems.ACM,2013.
[13]ABDRABBAH S B,AYADI M,AYACHI R,et al.Aggregating Top-K lists in group recommendation using borda rule[J].2017:325-334.
[14]DEY A K,ABOWD G D.Towards a better understanding of context and context-awareness[J].Lecture Notes In Computer Science,1999(1707):304-307.
[15]BERKOVSKY S,BERKOVSKY S,LUCA E W D.Group recommendation in context[C]//Challenge on Context-Aware Movie Recommendation.ACM,2011:2-4.
[16]LAI C H,HONG P R.Group recommendation based on the analysis of group influence and review content[C]//Asian Conference on Intelligent Information and Database Systems.Springer,Cham,2017:100-109.
[17]CARALIS J,KOGAN N,NAKAMURA M,et al.Systems and methods for generating location-based group recommendations[J].2017,32(5):115-120.
[18]KHOSHKANGINI R,PINI M S,ROSSI F.A Self-adaptive context-aware group recommender system[M]//AI*IA 2016Advances in Artificial Intelligence.Springer International Publishing,2017.
[19]QUIJANO-SANCHEZ L,SAUER C,RECIO-GARCIA J A,et al.Make it personal:a social explanation system applied to group recommendations[J].Expert Systems with Applications,2017,76:36-48.
[20]RECALDE L.A Social Framework for set recommendation in group recommender systems[C]//European Conference on Information Retrieval.Springer,Cham,2017:735-743.
[21]CANTADOR I,KONSTAS I,JOSE J M.Categorising social tags to improve folksonomy-based recommendations[J].Web Semantics Science Services&Agents on the World Wide Web,2011,9(1):1-15.
[22]张有志,王军.基于Folksonomy的本体构建探索[J].图书情报工作,2008(12):122-125.
[23]何金晶.基于社会化标签的本体构建与进化研究[D].西安:西安电子科技大学,2014.
[24]WU Z,PALMER M.Verb Semantics and lexical selection[J].Acl Proceedings of Annual Meeting on Association for Computational Linguistics,1994:133-138.
[25]LEE S,SEO W,KANG D,et al.A framework for supporting bottom-up ontology evolution for discovery and description of Grid services[J].Expert Systems with Applications,2007,32(2):376-385.