基于不确定性推理的家庭数据融合算法研究

Research of Family Data Fusion Pattern Based on Uncertainty Reasoning

摘要: 现有的数据融合模式往往无法将同一数据库中具有家庭关系的个体关联起来，使得目标客户的选择存在重复选取的局限性，同时数据属性的缺失也为进一步的决策带来困难。首先从不确定性推理模型出发，设计出用户兴趣、所处行业、产品偏好等的可信度推理算法，利用该算法将个性化数据库、杂志订阅数据库进行有效融合，在完善关键数据属性的基础上，生成新的标准化数据库；以年龄、性别、姓氏等属性为依据，制定家庭结构的识别规则，从而实现家庭结构数据的融合。利用邮政行业数据及自建数据库进行实验和分析，证明了方法的可行性和有效性。

Abstract: Existing data mining models are often unable to associate individuals with family relationships in the same database. The interest inference algorithm, industry inference algorithm and production inference algorithm are proposed based on certainty factor. Personalized Database and magazine subscription database were fused effectively, and generate new standardized database was generated with key data attributes. An identification rule of family structure was developed as the basis of age, gender, surname and other property, and the integration of family structure data was achieved. Experiments and analysis demonstrated the feasibility and effectiveness.