Automated contract review has limitations in adapting to diverse clauses dynamically and executing complex review rules accurately. Therefore， a Dynamic Rule generation and Neuro-Symbolic reasoning framework for contract Review （DRNS-Review） using Large Language Models （LLMs） was proposed. First， a dynamic rule generation method was designed， domain knowledge and contract context were integrated， and LLMs were used to generate natural language and symbolic review rules in real time. Then， a contract review mechanism based on neuro-symbolic reasoning was constructed to extract local facts automatically， rules were localized through symbolic matching， and explainable reasoning was completed by combining symbolic reduction with LLM semantic execution. Finally， reasoning outputs from multiple segments were aggregated， and a dual?track review report， comprising both structured and natural?language content， was generated. Experimental results demonstrate that DRNS-Review improves the risk identification F1 value by 13.0 and 8.3 percentage points， respectively， and improves the Risk Level Accuracy （RLA） by 13.3 and 8.7 percentage points， respectively， compared to baseline methods LLM-Direct and ManualRule-LLM. It can be seen that integrating dynamic rule generation with neuro-symbolic reasoning improves the accuracy and adaptability of contract review effectively， providing a generalizable technical pathway for intelligent compliance applications.

The core idea of prompt tuning is to insert a prompt template into the original input text and convert the classification problem into a language model containing mask code to predict the occurrence probability of the mask code in a sentence. In Aspect-Based Sentiment Analysis （ABSA） tasks， determining appropriate prompt templates requires corresponding syntactic knowledge， and constructing effective sentiment labels is very time-consuming. Furthermore， the semantic information and priori knowledge contained in complex implicit sentimental opinions cannot be ignored. Therefore， a syntactic prompt template was introduced， which incorporated syntactic knowledge （such as phrase structure and dependency relations） into the prompt templates for sentimental opinion mining related to sentiment aspect words to enhance the capture of explicit or implicit sentiment relation pairs. Experimental results on four publicly available datasets show that the proposed syntactic knowledge fused prompt model， SynPrompt （Syntax aware Prompt-tuning）， is effective； and compared to dotGCN （discrete opinion tree Graph Convolutional Network），DualGCN （Dual Graph Convolutional Network）， and dotGCN models， it has improvements of 0.81%， 0.27%， and 0.09% in accuracy， respectively. In addition， results of ablation experiments and case studies demonstrate that syntactic knowledge is effective on both explicit and implicit sentimental prompts. With 8-shot， by means of data augmentation， the F1 scores of SynPrompt model are improved by 31.62%， 42.02%， 121.04%， and 35.01% on the four datasets， respectively. However， with 16 and 32-shot， the accuracies of SynPrompt model are not improved significantly， indicating that the data augmentation approach is effective on few-shot datasets and enhances the ability of SynPrompt model to capture information.