Abstarct: This research, entixtled "Opinion Mining baxsed on Aspect, Category, Opinion and Sentiment Detection," presents a novel method for extracting Aspect-Category-Opinion-Sentiment quadruples (ACOS) with a special focus on implicit sentiment analysis. The proposed method is developed baxsed on the T5-large model and a distance-baxsed approach for ACOS extraction and classification. To address the data scarcity challenge, data augmentation techniques using synthetic sentence generation by ChatGPT and transfer learning from the Laptop-ACOS dataset were employed. Furthermore, techniques such as constrained decoding, advanced beam search (with 5 beams), multi-path decoding with voting aggregation, and integration of distance information and an additional attention mechanism were applied. Experiments were conducted on the REST16 dataset from SemEval-2016, and the results show a Precision of 85.20%, Recall of 80.30%, and F1-score of 82.00%, which represents a significant improvement over previous methods such as Double-Propagation-ACOS, JET-BERT-ACOS, and Distance-Extract-Classify-ACOS (with a maximum F1-score of 45.99%). This method demonstrates better performance in identifying implicit aspects and opinions, and effectively addresses the challenges of unified modeling and data scarcity.