Abstract:

The Internet is Web 3.0 era of knowledge connectivity, and its goal is a smarter network that can be understood by both the people and the machines. Based on this background, various types of knowledge graph emerge. The product knowledge is more challenging because of the heterogeneity of knowledge. This paper first designs a large-scale product knowledge organization framework that combines the objective knowledge of product categories with the subjective users’ opinions. Then, a learned index based on neural network is proposed to improve the query efficiency. Finally, a kind of sub variable joint strategy is realized based on the characteristics of the product knowledge structure and query requirements. The experimental results show that compared with the existing knowledge management system, the method proposed in this paper has greatly improved the retrieval efficiency of large-scale good knowledge.

Key words: product knowledge graph, knowledge organization, learned index, query optimization

摘要：

互联网正面向知识互联的Web3.0时代，其目标是实现人和机器都可以理解的更智能化的网络。在此环境下，各种类型的知识图谱应运而生。商品知识由于知识的异质性，使其管理更具挑战性。设计一种融合了客观性商品分类知识和主观性用户观点的大规模商品知识组织框架；提出了一种基于神经网络的学习索引技术，以此提升查询效率；针对商品知识结构的特性和查询需求的特点，实现了一种基于子变量组合的连接策略。实验结果表明，提出的方法相对于现有的知识管理系统，在大规模商品知识的检索效率上有较大幅度的提升。

关键词: 商品知识图谱, 知识组织, 学习索引, 查询优化