社会化商务环境下基于中智集和云模型的推荐方法研究

In social commerce, the data of social network and transaction might show some features of big data that could be characterized by large quantity, high growth and diversity, which makes it a more difficult problem for consumers’ purchase decision-making. Thus it is a challenge for social commerce to construct an effective commodity recommendation method involved in enormous users and amount of commodities to provide effective support to purchase decision-making for consumers. For this reason, the project may focus on the commodity recommendation method for social commerce. Firstly, investigate and recognize the social factors which influence consumers’ decision-making and study the measurement approaches of social factors by using neutrosophic sets theory. And then, aggregation approaches for assessment information will be explored. Besides, mine and mark consumer’s group properties such as hobbies and relation, calculate the group similarity between consumers to discover similar group, on the basis of which, commodity recommendation method with cloud model will be proposed. For the recommendation method could reflect the consensus degree of similar group and it takes into account the social factors, consumers could make purchase decision effectively and accurately. Finally, the application of the method in real social business environment will be studied to verify its practicality and feasibility to provide a theoretical basis and practical guidance for the commodity recommendation in social commerce.

当前社会化商务环境下，社交网络数据和商品交易数据开始呈现体量大、高增长和多样化等大数据特征，使消费者购买决策更加困难。针对海量用户和海量商品提出有效的商品推荐方法，从而为消费者提供更精准的购买决策支持，是社会化商务发展面临的挑战。针对此，本项目拟以社会化商务环境下的商品推荐方法为研究对象，在识别影响消费者购买决策的社会化因素的基础上，采用中智集等理论对商品评价信息进行量化，研究各社会化因素下商品评价信息的集结方法，并对消费者的兴趣爱好、用户关系等群体属性进行挖掘和标注，计算消费者间的群体属性相似度，以识别相似群体，在此基础上提出基于云模型的商品推荐方法。该方法体现反映相似群体共识度的商品综合评价意见，并综合考虑社会化因素，以辅助消费者有效进行购买决策。此外，研究这些方法在实际社会化商务环境中的应用，验证这些方法的有效性和可行性，为社会化商务环境下的商品推荐提供理论依据和实践指导。

社会化商务背景下，在线评论为商品推荐提供了大量可参考的信息。这些信息的不确定性、模糊性、不完全性特征，使得传统的信息量化方法难以准确表达，因而降低了推荐方法的准确性。针对此，本项目以社会化商务环境下的商品推荐方法为研究对象，（1）研究了影响消费者购买决策因素的识别与量化方法。本项目提出了基于中智集及扩展中智集、基于概率语言术语集、基于犹豫模糊集及扩展犹豫模糊集的信息量化方法。（2）研究了社会化因素的权重确认方法与信息集结方法。本项目提出了基于最大离差法、熵权法的客观权重确认方法。针对社会化因素间交互与不交互情形，分别提出了基于模糊加权平均集结算子、模糊有序加权平均集结算子以及模糊优先集结算子的信息集结方法，与基于Bonferroni mean (BM) 算子的信息集结方法。（3）研究了社会化商务环境下的商品推荐方法。本项目提出了基于多准则决策的个性化推荐方法与基于多准则决策的群推荐方法。（4）为验证本项目提出的推荐方法有效性，本项目依托TripAdvisor、Tmall、PConline网站数据集进行实验，通过与传统方法对比，逐一验证了本项目提出方法的有效性与可行性。. 经过3年的努力工作，本项目全面完成了预定的研究内容。在国内外重要期刊及国际学术会议上发表了相关学术论文20篇，其中SCI期刊论文18篇，SSCI期刊论文1篇，EI检索会议论文1篇，ESI高被引6篇。