The current unstructured multi-source data fusion has insufficient application of data attributes, which leads to unsatisfactory data fusion results. To this end, an unstructured multi-source data fusion method based on knowledge graph is proposed. Building a data tracking model to integrate unstructured multi-source datasets. Processing unstructured multi-source datasets. Associating data based on knowledge graphs. Alignment and complementary processing of data attributes to correlate unstructured multi-source data attribute fusion. The experiments show that the average fusion of the method reaches 0.92152. The fusion of unstructured multi-source data is substantially improved and has high practical application value.