Government data resources are characterized by a wide range of sources, diverse types, large data volumes and unclear data distribution, which can lead to lacks of unified management, high efficiency in data access and use, and ability to continuously promote the data release value. In order to solve the above problems and realize the correlative fusion of data from different sources and different types, this paper adopts the fusion of multi-source heterogeneous data to provide technical means for the exchange and sharing of government data. This paper proposes the concept of information channel and channel coupling, and provides a multisource heterogeneous information channel coupling method, which includes three module for constructing the Initial channel coupling knowledge graph, realizing the channel data coupling based on the channel coupling knowledge graph, and realizing the knowledge updating based on the channel coupling knowledge graph, respectively. In the proposed method, the technologies of atlas construction, knowledge extraction, knowledge fusion, knowledge processing and knowledge updating are integrated together, and a top-down sorting retrieval model is proposed to accelerate the speed and accuracy of knowledge fusion and data retrieval.