论文部分内容阅读
【目的/意义】基于移动互联网的社交网络飞速发展,对传统的推荐技术提出了新挑战。为了开展进一步研究,厘清种类与方法十分必要。【方法/过程】分析了传统互联网推荐系统、信息物理方法以及机器学习方法三大类推荐系统。通过研究代表性的论文,比较了常见推荐算法模型的优缺点。【结果/结论】结果表明:很难找到一种“万能的”模型,使移动社交网络的推荐系统在实时性、准确性和多样性等关键特征取得满意的结果。移动社交网络推荐技术可能取得突破的领域,将是多种模型的综合运用。
[Purpose / Significance] The rapid development of social networks based on mobile Internet poses new challenges to traditional recommended technologies. In order to carry out further research, it is necessary to clarify the types and methods. 【Method / Process】 The paper analyzes three kinds of recommendation system of traditional Internet recommendation system, information physics method and machine learning method. By studying representative papers, the advantages and disadvantages of common recommended algorithm models are compared. [Results / Conclusion] The results show that it is very hard to find a “universal” model, which makes the recommendation system of mobile social network obtain satisfactory results in the key features such as real-time, accuracy and diversity. The areas where mobile social network recommendation technologies may make a breakthrough will be the combined use of multiple models.