With the rapid development of the digital economy, traffic has become the key to the profitability of self-media, and some self-media use script-washing software to generate pseudo-original content in batches to seek improper benefits, which has seriously damaged the content creation intellectual property ecology. Aiming at the problem that it is difficult to detect pseudo-original manuscripts, this article discusses from a technical level, proposes an anti-plagiarism technology based on the fusion of semantic similarity and emotional tendency, uses topic mining technology to obtain text topic distribution, calculate topic similarity, and uses text representation technology Calculate the semantic similarity of the text, use the sentiment analysis technology to calculate the difference in the emotional tendency of the text, and merge the topic similarity, semantic similarity and emotional tendency to detect plagiarism. The experiment was conducted by collecting data on a public platform. The results show that the accuracy of this method to judge plagiarism is 91.9%, which has certain practical significance in the technical practice of antiplagiarism.