基于机器学习的商品本体细粒度语义知识获取

Acquisition of Fine-granularity Semantic Knowledge of Commodity Ontology Using Machine Learning Methods

针对电子商务应用中商品本体模型粒度过粗和细粒度语义知识匮乏的问题,提出了商品候选属性集的5类分类特征,选择进化算法对分类特征集进行优化,研究基于机器学习的商品本体细粒度语义知识获取方法。通过SVM算法执行分类实验,结果证明了5类特征集的有效性。所提出的5类特征集对于其他领域具有一定的通用性,获取细粒度语义知识也有助于构建商品细粒度语义知识库,满足电子商务应用中对细粒度商品知识的需求。

When applied to E -commerce domain, the present commodity ontology model has two problems： the granularity of ontology model being too coarse and its fine - granularity semantic knowledge being too scarce. So five types of classification characters were proposed from commodity＇s candidate attributes. An evolutionary algorithm was chosen to optimize the classifica- tion character set. And a method to gain fine - granularity semantic knowledge was studied using supervised learning methods. The classification experiments proved the validity of the five types of character set with SVM algorithms. The proposed classifica- tion characters have a certain commonality in other areas, and the method can help to build fine - granularity semantic knowledge base of commodity, which can meet the needs of fine -granularity commodity knowledge in E -commerce field.