Science and technology writing recommendation method based on self-attention mechanism

The invention relates to a science and technology writing recommendation method based on a self-attention mechanism, and belongs to the technical field of computer natural language processing and social media. The science and technology writing recommendation method comprises the steps: firstly, utilizing existing paragraph writing content to mine writing themes and intentions of an author, and preliminarily screening candidate paragraphs according to the writing themes and intentions; secondly, designing a deep science and technology writing recommendation model based on a self-attention mechanism to obtain deep semantic representation of a context so as to sort candidate paragraphs; and comparing the prediction sorting result with a real sorting result to obtain a recall rate and a normalized loss accumulation gain so as to evaluate the quality of the writing recommendation method. Compared with a traditional content-based citation recommendation method, the method has the advantagesthat recommendation pred