The invention provides an automatic text abstracting method and device based on global semantics, a medium and equipment. The method comprises the following steps: preprocessing the content of an original text, dividing the content according to bytes, and replacing capital letters in the original text with lowercase letters to obtain text information; by an encoder, performing global encoding on the text information based on a convolutional neural network and a self-attention mechanism, and performing screening through a control unit to obtain a final encoding output result; and by the decoder, decoding the encoded output result based on a repetitive punishment mechanism and generating a text summary. According to the method, a convolution filter is added to athe encoder, and the repetitive punishment mechanism is used in the decoder part to further suppress repetitive words so that the smoothness of abstract text semantics can be improved, the repeated punishment mechanism can greatlyinhibit already appeared words, the repeatability problem of abstract generation is reduced, and therefore the readability of abstract generation is improved.