25-08-2023 дата публикации
Номер: CN116644177A
Принадлежит:
The invention provides a text abstract generation method based on global autoregression, which comprises the following steps: S1, inputting a text into a first encoder to obtain semantic representation of the text; s2, inputting the semantic representation H of the text into a decoder to generate candidate abstracts; s3, inputting the candidate abstract and the text semantic representation H into a second encoder to obtain a perceptual semantic representation H of the candidate abstract; s4, inputting the text semantic representation H and the perceptual semantic representation H of the candidate abstract into a decoder to generate a new candidate abstract; and S5, repeating the steps S3-S4 until a preset number of repetition times is reached, and finally taking the candidate abstract output by the decoder as the text abstract. Through the global autoregression method provided by the invention, the generated text can be repeatedly iterated and rewritten, the content of the output text is ...
Подробнее