Abstract: 【Purpose/significance】With the popularity of online shopping, online reviews have become important data that affect the de-
cision-making of consumers, sellers and producers. In the era of big data, online reviews show the characteristics of multi-source het-
erogeneous and explosive growth, and it is difficult to provide strong intelligence support for users' purchasing decisions and merchant
competition.【Method/process】Therefore, this article proposes a framework for constructing an online comment knowledge graph
based on multiple data sources. The model layer is constructed from top to bottom around the source, content, and form characteristics of online reviews to form the conceptual framework of the knowledge graph. Obtain entities, relationships and attributes from multi-source heterogeneous texts, and perform data fusion and knowledge graph analysis.【Result/conclusion】The experimental part takes the online reviews of mobile products as an example to verify the effectiveness of the knowledge graph constructed in this article for on-line review mining and analysis. The research results reveal the characteristics of multi-source heterogeneous online review data, and provide a new research perspective for the organization, display and mining of online review information in a big data environment.【In-novation/limitation】Use knowledge graphs to describe online reviews, effectively solving problems such as information overload and multi-source heterogeneous information fusion. This article uses a semi-automated method to construct the knowledge graph, and in the future, we will consider introducing an unsupervised method to improve the construction efficiency.