Enterprises operating across borders often face conflicting data protection regulations and must address coordinated execution across heterogeneous blockchains， making compliant cross-border data transfer complex and risky. Therefore， a smart contract generation framework driven by Large Language Model （LLM） and Semantic Web Rule Language （SWRL） was proposed for cross-border data compliance across heterogeneous blockchains. First， centering on formalized rule definition-driven access control， cross-chain interoperability between two heterogeneous blockchains was achieved by cooperating with notarization services. Additionally， the data subject's latest authorization was injected into access decisions in real-time by utilizing an integrated consent module. Second， a rule matching framework based on finite-state decision process was constructed to decompose Semantic Web Rule Language （SWRL） compliance rules into reusable and modular logical units. Finally， a fine-tuned LLM was introduced to complete the SWRL rules semantically， which were automatically converted into executable smart contract code on the blockchain. Experimental results show that in the smart contract code generation task， the fine-tuned LLM achieves a comprehensive code generation success rate （FullRate） of 76.1%， which is improved by 18.6， 3.2， and 1.2 percentage points compared to that of the baseline models LLaMA-3-8B-Instruct， DeepSeek-V3， and DeepSeek-R1， respectively. Furthermore， the performance of the prototype system was evaluated under varying network node counts and adjustable channel configurations， demonstrating that it meets enterprise-level scalability requirements. The proposed framework significantly enhances the efficiency of smart contract generation while maintaining high security and logical correctness， demonstrating the practical feasibility and engineering application value of the framework in efficient and compliant cross-border data sharing scenarios.

Aiming at the problems such as insufficient inter-domain collaboration ability， heterogeneous resource identifiers and low detection efficiency in conflict detection of cross-domain defense strategies in programmable networks， a cross-domain defense strategy conflict detection method based on multi-domain policy graph was proposed. Firstly， an intent-driven defense strategy model was constructed on the basis of the general JSON language， and the precise association between defense intentions and defense strategies was achieved through semantic label injection， so as to solve closed problem of the single-domain strategy model. Secondly， the Layered Hash Mapping （LHM） algorithm was utilized to generate Global Resource Identifier （GRI）， thereby solving the problem of resource identifier conflicts in multi-controller domains. Finally， by constructing a Multi-Domain Joint Policy Graph （MD-JPG）， as well as integrating the topological， action and resource dependencies among cross-domain strategies， a Cross-domain Conflict Detection algorithm based on graph traversal in four Dimensions， （CDC-4D） was designed to identify action conflicts， rule coverage conflicts， resource competition conflicts and strategy type conflicts accurately. Experimental results show that in the multi-controller network defense scenarios， the strategy conflict detection latency， memory usage， and detection F1-score of the proposed method have achieved good results.

The speech signals were analyzed by means of shorttime Fourier transform, and the time series clusters that were composed of the same frequency's energy series of each speech were gained. Between every column vector serie and the serie consisting of every row vector's average value, the linear regression equations were established based on time series pretreatment and mathematical statistics. The deterministic parts and stochastic parts of the time series cluster were separated, and the feature parameters were extracted from speech signals of the speaker, then the speech signals were recognized. The experimental results show that the highest average recognition rate is 97.94%, with the distance parameter described in this paper and three frequencies on the speech set of 194 speech signals pronounced by eight speakers.

Based on the characteristics of the Chinese named entity relation extraction, features were selected and feature vectors were constructed in terms of Chinese morphological, grammar and semantics. Then potential named entity pairs in accordance with the specific entity relation template were extracted and divided into positive and negative cases. Support Vector Machine (SVM) classifier was trained by the positive and negative cases and used to judge the relation of the potential named entity pairs. Experimental results prove that this new method can effectively improve the accuracy of Chinese named entity relation extraction.