Dialogue Text Summarization Generation Based on Multiple Features Fusion Filtering
A lot of useless information in the original dialogue can interfere with the model's attention to important infor-mation.To this end,a dialogue summarization model based on multiple feature fusion filtering is proposed,which can adaptively integrate multiple semantic features to filter out useless information and achieve more accurate summary genera-tion.The experimental results on the dialogue summarization dataset CSDS show that compared with advanced models such as BART,MV-BART,and BART(DALL),this method can improve the ROUGE score by up to 2.89％.
dialogue summarizationtext summarizationmultiple features fusionBART
万方数据
维普
