System log anomaly detection method and device, equipment and medium

The invention discloses a system log anomaly detection method and device, equipment and a medium, and belongs to the technical field of system logs. Semantic analysis is carried out on the log text, and near and antonym pairs in the log text are extracted; performing template analysis on the log text based on the word pair information of the near and antonym word pairs to generate a log template; the sentence vector generation model is finely tuned based on the synonym and antonym pairs, and sentence vectors corresponding to the log template are generated by using the finely tuned sentence vector generation model; and initializing a preset prediction model based on the sentence vector, and detecting abnormity in the log text according to the initialized prediction model. Whether the system logs are abnormal or not is comprehensively judged by analyzing the semantic features and the sequence features of the system logs. Therefore, the prediction model can pay attention to the semantic and sequence information