Abstract: According to attention mechanism-based encoder-decoder model for data-to-text generation does not fully consider the structure information of the slot-value pair data and the integrity of ...