This paper focuses on resource-constrained scenarios and aims to build a lightweight intelligent dialogue system, balancing the depth of intent understanding, the quality of response generation, and the computational overhead, to enhance the accuracy of task-oriented dialogues and the fluency of open-domain dialogues. The core innovations include three aspects: First, the Semantic Entity Enhanced Intent Recognition (SEEIR) model is designed, which separates the text backbone from the entity information, introduces a collaborative interaction mechanism, auxiliary learning tasks, and attention isolation strategies, to enhance the perception and utilization of key semantic entities, improving the accuracy and generalization ability of intent classification in complex situations; Second, a hierarchical hybrid generation architecture is constructed, integrating precise responses based on rules, high-quality examples based on retrieval, and general responses based on generation. Through efficient decision paths for flexible invocation, it ensures the accuracy and diversity of responses while reducing computational dependence and overcoming the bottleneck of high-concurrency deployment; Third, a few-shot chained prompt data augmentation method is proposed, relying on the self-generation of high-quality and diverse training data by large models, to solve the data scarcity problem during the cold-start stage and help the system quickly adapt to new domains. Based on the above solutions, the system has been implemented. Experiments show that this system outperforms traditional baseline models in intent recognition accuracy and response generation quality on public datasets and self-built business datasets, and the response speed and resource consumption meet the requirements of lightweight deployment, providing a practical and feasible solution for the application of intelligent dialogue technology in low-cost and high-concurrency scenarios.

Qizeng Sun. Research on Lightweight Intelligent Dialogue Systems Based on Semantic Entity Enhanced Intention Recognition and Rule Retrieval Generation Hybrid Models. International Journal of Big Data Intelligent Technology (2026), Vol. 7, Issue 1: 60-70. https://doi.org/10.38007/IJBDIT.2026.070108.

[1] Deriu J, Rodrigo A, Otegi A, et al. Survey on evaluation methods for dialogue systems. Artificial Intelligence Review, 2021, 54(1): 755-810.

[2] Ni J, Young T, Pandelea V, et al. Recent advances in deep learning based dialogue systems: A systematic survey. Artificial intelligence review, 2023, 56(4): 3055-3155.

[3] Huang, J. (2025). Research on Cloud Computing Resource Scheduling Strategy Based on Big Data and Machine Learning. European Journal of Business, Economics & Management, 1(3), 104-110.

[4] Xu H, Zhang H, Lin T E. Intent Recognition[M]//Intent Recognition for Human-Machine Interactions. Springer Nature Singapore, 2023: 7-29.

[5] Liu X. The Role of Generative AI in the Evolution of Digital Advertising Products. Journal of Media, Journalism & Communication Studies, 2025, 1(1): 48-55.

[6] Bai J, Yan Z, Zhang S, et al. Infusing internalized knowledge of language models into hybrid prompts for knowledgeable dialogue generation. Knowledge-Based Systems, 2024, 296: 111874.

[7] Zhou, Y. (2025). Improvement of Advertising Data Processing Efficiency Through Anomaly Detection and Recovery Mechanism. Journal of Media, Journalism & Communication Studies, 1(1), 80-86.

[8] Huijie Pan. Design of Data-Driven Social Network Platforms and Optimization of Big Data Analysis. Machine Learning Theory and Practice (2025), Vol. 5, Issue 1: 133-140.

[9] Yixian Jiang. Research on Integration and Optimization Strategies of Cross-platform Machine Learning Services. Machine Learning Theory and Practice (2025), Vol. 5, Issue 1: 141-148.

[10] Jiangnan Huang. Application of AI-driven Personalized Recommendation Technology in E-commerce. International Journal of Neural Network (2025), Vol. 4, Issue 1: 40-47.

[11] Huijie Pan. Discussion on Low-Latency Computing Strategies in Real-Time Hardware Generation. International Journal of Neural Network (2025), Vol. 4, Issue 1: 48-56.

[12] Huijie Pan. Discussion on Low-Latency Computing Strategies in Real-Time Hardware Generation. International Journal of Neural Network (2025), Vol. 4, Issue 1: 57-64.

[13] Chuying Lu. Object Detection and Image Segmentation Algorithm Optimization in High-Resolution Remote Sensing Images. International Journal of Multimedia Computing (2025), Vol. 6, Issue 1: 144-151.

[14] Buqin Wang. Research on Load Balancing Technology in Distributed System Architecture. International Journal of Multimedia Computing (2025), Vol. 6, Issue 1: 152-159.

[15] Yajing Cai. Distributed Architecture and Performance Optimization for Smart Device Management. International Journal of Big Data Intelligent Technology (2025), Vol. 6, Issue 2: 130-138.