<p>Aspect Sentiment Triple Extraction (ASTE) is a fine-grained aspect-level sentiment analysis aimed at extracting aspect words, opinion words, and the sentiment polarity between them in text. Existing end-to-end ASTE models tend to ignore information such as the potential lexical relationships between words and their differences in importance, and fail to take full advantage of the interaction between lexical features and dependency features. To address the above problem, we propose a Quantification of Part-of-speech Relationships for Aspect Sentiment Triplet Extraction (QoPR-ASTE). The model first extracts contextual semantic features by combinatorial encoder; then analyzes the connection and dependency between lexical pairs of words to mine potential syntactic information, and assigns weights according to the importance, constructs lexical adjacency weight matrices and dependency type weight matrices, so as to remove the redundant information and strengthen the expression of syntactic features; then extracts the syntactic information by the two-group graphical convolutional network and the syntactic interaction module and fusion; finally, the semantic and syntactic information is fused through the multi-head attention graph transformation module, and the sentence is labeled using word pair relations and decoded to extract the triad. The F1 values of this model on the four public datasets demonstrate an improvement of 0.80%, 1.46%, 0.92%, and 0.78% compared to the PBLUN model, respectively. The experimental results indicate that the model is capable of efficiently mining potential semantic and syntactic information, thereby significantly enhancing the accuracy of triplet extraction.</p>