Electric power multi-modal entity relationship extraction method and device based on transfer learning

The invention relates to a transfer learning-based electric power multi-modal entity relationship extraction method. The method comprises the following steps of collecting and preprocessing electric power text and image data; constructing a cross-modal relation feature extraction model to obtain an advanced feature rt of the multi-modal data in the power field and an advanced feature rs of the multi-modal corpus in the open field; constructing a transfer learning-based power multi-modal entity relationship extraction model: adapting the high-level features rs to the high-level features rt corresponding to the power field to minimize the difference value between the high-level features rt and rs, and inputting the high-level features rt and rs with the difference value smaller than a preset threshold value as results to the next layer; and the classification relation output layer outputs a relation identification result. The method has the beneficial effects that rich corpora in other fields are utilized to ex