The invention discloses a data fusion method and device, belonging to the technical field of computers. A specific embodiment of the method comprises the following steps of: screening and combining original data tables of various data sources to obtain an associated data table of each data source, and adding a dimension tag to a field in the associated data table; carrying out feature analysis onfields in the associated data table, and selecting features to be used; extracting gradients from the associated data tables according to the dimension tags and the standby features; gradient-based fusion model. According to the embodiment, the data of different data sources can be collected and efficiently processed, the data of different structures can be safely and rapidly fused, and the data with the same standard and format can be provided for constructing a unified fusion information scoring system.